INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Human herpesvirus type 8 (HHV8) is the second human herpesvirus classified in the gamma subfamily. HHV8 has been causally associated with Kaposi's sarcoma (KS) (6, 31), body cavity-based lymphoma (BCBL), also known as primary effusion lymphoma (5), and multicentric Castleman's disease (45). Though the virus is in a tightly latent state in the majority of infected cells (43, 51), it has been suggested that in KS lesions, lytic cycle gene products may be involved in the proliferation of neighboring cells, contributing to pathogenesis (46). A reduction in the number of virions would both directly and indirectly decrease the likelihood of transmission to cellular targets with the capacity to proliferate uncontrollably upon infection. Prophylactic antiviral therapy to limit in vivo lytic HHV8 replication, particularly in the T-cell-deficient host, may decrease the incidence of KS and other virus-associated diseases. Therefore, conventional antiviral strategies are being examined, both retrospectively and prospectively, as a way to prevent development of HHV8-associated diseases (2, 9, 14, 20, 21, 28, 39).

A few studies have examined the effect of a limited number of antiviral drugs on HHV8 replication in vitro (22, 29, 33). Of the nucleosides that require a viral thymidine kinase (TK) for activation, ganciclovir (GCV), an acyclic guanine analog, was the most effective at inhibiting viral replication, while penciclovir (PCV) and acyclovir (ACV), also acyclic guanine analogs, and the thymidine analog (E)-5-(2-bromovinyl)-2'-deoxyuridine (BVDU) were weakly or not at all effective. A thorough examination of the substrate specificity of the virus-encoded TK will be important in determining which drugs may be active in inhibiting viral replication. Moore et al. identified open reading frame (ORF) 21 in the HHV8 genome as encoding the putative TK based on similarity to Epstein-Barr virus (EBV) and herpesvirus saimiri (HVS) TKs (32). Although a recent study suggests that heterologous expression of HHV8 TK can confer sensitivity to GCV (4), its reactivity toward other nucleosides has not yet been examined, nor has any TK activity been demonstrated.

We investigated the enzymatic activity of ORF 21 by expression in TK mammalian cells and as a purified glutathione S-transferase (GST) fusion protein from bacteria. Results indicate that the protein encoded by ORF 21 does have TK activity, albeit relatively low, and has a higher K_m for dThd compared to other human herpesvirus TKs. Human TK cells expressing HHV8 TK are not sensitized to GCV, which is found to be metabolized very poorly in these cells, and purified GST-HHV8 TK does not phosphorylate GCV. In contrast, the enzyme efficiently phosphorylates zidovudine (AZT) to AZT 5'-diphosphate (AZT-DP). Its reactivity toward a panel of antiviral nucleoside analogs indicates that it has a limited substrate specificity similar to other gammaherpesviruses and suggest modifications which may be useful in designing additional nucleoside analogs to target this enzyme.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Nucleosides and cells. Tritium-labeled dThd (6.7 Ci/mmol) was obtained from New England Nuclear, Boston, Mass. Tritium-labeled AZT (11.7 Ci/mmol), tritium-labeled GCV (12.4 Ci/mmol), and AZT-DP were obtained from Moravek Biochemicals, Brea, Calif. The chemical purities of [³H]GCV, [³H]AZT, and AZT-DP were 99.8, 99.8, and 98.8%, respectively, as determined by the manufacturer. The unlabeled nucleosides dThd, 2'-deoxyguanosine (dGuo), 2'-deoxyadenosine (dAdo), 2'-deoxycytidine (dCyd), 5-bromo-2'-deoxyuridine (BrdU), 5-iododeoxyuridine (IdU), BVDU, 3'-deoxy-2',3'-didehydrothymidine (D4T), and AZT were from Sigma, St. Louis, Mo. 2'-Deoxy-2',2'-difluorocytidine (Gemcitibine, dFdC) was from Eli Lilly, Indianapolis, Ind.; 1--D-arabinofuranosyl-E-5-(2-bromovinyl)uracil (BVara-U) was from Bristol-Myers Squibb, Princeton, N.J. GCV was from Roche Laboratories, Palo Alto, Calif., ACV was from Glaxo Wellcome, Research Triangle Park, N.C., and the cytidine nucleotide analog (5)-l-[3-hydroxy-2-(phosphonylmethoxy)propyl]cytosine (HPMPC) or Cidofovir, (CDV) was obtained from Gilead, Foster City, Calif. 2'-Deoxy-2'-fluoro-arabinofuranosyl-5-iodouridine (FIAU), D- and L-2'-deoxy-2'-fluoro-arabinofuranosyl-5-methyluracil (D- and L-FMAU), 2'-deoxy-2'-fluoroarabinofuranosyl-5-ethyluracil (FEAU), 5-ethyl-d-U (EtdU), 5-bromouridine arabinofuranoside (ara-BrU), 5-ethynyl-2'-deoxyuridine, D- and L-5-iodouridine arabinofuranoside (D- and L-ara-IU), thymine arabinofuranoside (ara-T), D- and L-uridine arabinofuranoside (D- and L-ara-U), PCV, buciclovir (BCV), and 2'-deoxy-2'-fluoro-arabinofuranosyl-5-methylcytidine (FMAC) were chemically synthesized by R. Schinazi. All unlabeled nucleoside and nucleotides were stored as concentrated stock solutions at 20°C.

Construction of TK vectors. pML18 and pML21, plasmids which contain partial genomic clones of HHV8 DNA from KS lung tumor and BCBL-1 cells (38), respectively, were a gift from Don Ganem, University of California, San Francisco. ORF 21 from both genomic sources was sequenced and found to be identical to the sequence of ORF 21 reported by Russo et al. (41) and obtained from the BC-1 cell line, except at position 35598, where both clones pML18 and pML21 show a TC change that conserves amino acid 72 as Thr. ORF 21, which encodes the putative HHV8 TK, was excised from pML18 by using the restriction enzymes ApaI and NsiI. The resulting 1,821-bp band was blunt-end cloned into the expression vector pCMV (7) at the NotI site by using established methods (42). Recombinants with the HHV8 TK gene in the correct orientation were identified by appropriate restriction enzyme digestion analysis. The gene was sequenced in its entirety, and the resulting construct was designated pCMV-HHV8-TK. The vector pCMV-EBV-TK was constructed as described elsewhere (15).

Expression and purification of GST fusion proteins. pGEX6P2-HHV8-TK and pGEX6P2 were used to transform competent Escherichia coli BL21 cells. One-liter 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 24 h at 20°C. Cells were pelleted at 5,000 × g in a J2-21 centrifuge (Beckman, Palo Alto, Calif.) and resuspended in 40 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 in a 550 Sonic Dismembrator (Fisher Scientific, Pittsburgh, Pa.). Clarified lysates were pelleted 30 min at 20,000 × g at 4°C. One milliliter of a 50% slurry of glutathione-Sepharose 4B (Pharmacia) was added to the clarified supernatants and incubated at ambient temperature with gentle rocking for 30 min. The Sepharose beads were pelleted at 500 × g in a Beckman GPR centrifuge and washed five times by resuspending beads in 40 ml of wash buffer (50 mM glucose, 25 mM Tris [pH 8], 10 mM EDTA, 5 mM 2-mercaptoethanol, 1 mM PMSF) and pelleting as described above. The GST fusion protein was eluted by incubating washed beads in 1 ml of elution buffer (wash buffer plus 10 mM reduced glutathione [Sigma]) for 10 min at ambient temperature. The eluted protein was divided into 50-µl aliquots and stored at 80°C. Proteins were quantitated by the method of Bradford (3), and protein purity was assessed by polyacrylamide gel electrophoresis (PAGE) of 30 µg on a sodium dodecyl sulfate (SDS)-10% polyacrylamide reducing gel followed by Coomassie blue staining. GST-EBV TK and GST-HSV1 TK were prepared as described in reference 15.

Protease cleavage of GST fusion. PreScission protease (Pharmacia) was used to cleave the GST moiety from the GST-HHV8 TK fusion protein. The protein was expressed in bacteria and harvested as described above except that following incubation of clarified supernatant with glutathione-Sepharose 4B, the beads were pelleted and washed once with 5 ml of cleavage buffer (50 mM Tris [pH 7.0], 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol [DTT]). The beads were then resuspended in 225 µl of cleavage buffer containing 36 U of PreScission protease and incubated at 4°C for 5 h. Following centrifugation, the HHV8 TK-containing supernatant was collected, quantitated, and analyzed as described above. GST and PreScission protease remain bound to the beads in this procedure.

Selection of TK-expressing cell lines. 143B TK cells were transfected with equimolar amounts of either pCMV, pCMV-HHV8-TK, or pCMV-EBV-TK, using lipofectAMINE (Life Technologies) according to the manufacturer's directions. Forty-eight hours after transfection, the cells were split 1:10 into DMEM-10 supplemented with 1× HAT to select for TK⁺ cells. Colonies were isolated and expanded and were designated 143B HHV8-TK or 143B EBV-TK cells. RNA blot analysis was used to confirm expression of TK RNA in the individual clones as described below. After separation of 143B HHV8-TK colonies into individual cultures, the medium was supplemented with 84 µM dThd (100 µM, final concentration) to reduce cell doubling time. Northern blot analysis was used to confirm expression of HHV8 TK RNA by using a previously described method (12). The probe was a 20-bp oligonucleotide, antisense to genome positions 35480 to 35499 and 3' end labeled with [³²P]dAdo (6,000 Ci/mmol; NEN), using a DNA tailing kit (Boehringer Mannheim, Indianapolis, Ind.) according to the manufacturer's instructions. Transcript size was estimated by extrapolation based upon the distance migrated by the 28S and 18S rRNAs and their reported sizes of 4,718 bases and 1,874 bases, respectively (25).

[³H]dThd incorporation into cellular DNA. Cells were split into 12-well plates in DMEM-10 (no HAT) containing 1 µM [³H]dThd. At 24, 48, and 72 h, the cells were washed, scraped into phosphate-buffered saline, pelleted, resuspended in 50 µl of lysis buffer (1% NP-40 in phosphate-buffered saline), and frozen at 80°C. After all time points were taken, samples were thawed on ice and 10 µl of each was spotted onto GF/C filters (Whatman, Maidstone, England). High-molecular-weight nucleic acid was precipitated on the filters by being washed three times in ice-cold 5% trichloroacetic acid (TCA)-20 mM sodium pyrophosphate (Sigma) and once in 70% ethanol. Filters were dried, placed in scintillation vials, and counted in the scintillation counter. Protein content in each lysate was determined by the method of Bradford (3). Results were expressed as counts per minute per milligram of protein.

Phosphorylation assay, buffer optimization, and K_m determination. Optimal conditions for pH, divalent cation usage and concentration, and ATP concentration for GST-HHV8 TK were determined in disc phosphorylation assays by varying amounts of the indicated reagents in an initial assay mixture containing 140 mM Tris (pH 7.5), 1.7 mM DTT, 8 mM NaF, 2 mM ATP, 2 mM MgCl₂, 5 mM PMSF, 30 µM [³H]dThd, and 80 µg of GST-HHV8 TK in a 100-µl final volume at 37°C. Twenty-microliter aliquots taken at 0, 20, 40, and 60 min were spotted onto DE-81 discs (Whatman) and washed four times for 15 min in 5 mM ammonium formate and one time for 15 min in 95% ethanol. The discs were dried, and the amount of radioactivity was determined by counting in a model LS 6500 multipurpose scintillation counter (Beckman). Results show a pH optimum of 7.5, a divalent cation preference for magnesium at 9 mM and a broad optimal ATP concentration ranging from 0.1 to 4 mM. ATP was found to be inhibitory above 4 mM in our system. All subsequent GST-HHV8 TK phosphorylation assays were performed as described above, using 10 µg of enzyme in 140 mM Tris (pH 7.5)-1.7 mM DTT-8 mM NaF-0.5 mM ATP-9 mM MgCl₂-5 mM PMSF.

Competition assay. To determine the relative binding of nucleoside analogs to HHV8 TK, cold nucleoside analogs in a 10-fold molar excess over dThd were added to phosphorylation reactions to compete with dThd. Briefly, 10 µg of GST-HHV8 TK was added to the phosphorylation assay buffer containing 30 µM [³H]dThd and 0.3 mM cold nucleoside analog in a 100-µl final volume. After incubation at 37°C for 60 min, the reactions were stopped and analyzed by disc assay as described above except that 50 µl of each reaction was spotted onto DE-81 discs. Results are reported as percent control of the reaction with dThd only.

Growth of cells in GCV. 143B HHV8-TK, 143B EBV-TK, and 143B HSV1-TK cells (15) were plated into 12-well dishes at a density of 5 × 10⁴ cells per well in DMEM-10-1× HAT medium containing 84 µM supplemental dThd. 143B TK cells were plated in the same manner except the medium was DMEM-10 (no HAT supplement) with 84 µM added dThd. The next day, fresh medium containing GCV ranging from 0 to 100 µM was placed on the cells in triplicate. Forty-eight hours later, wells were replenished with fresh medium containing the drug. On day 5, cells were trypsinized, resuspended in DMEM-10, and enumerated in the presence of trypan blue. Data are presented as percent viable cells compared to untreated control at each GCV concentration.

GCV metabolism. 143B TK, 143B HHV8-TK, 143B EBV-TK, and 143B HSV-TK cells were seeded at 10⁶ cells/well in six-well plates. Twenty-four hours later, the cells were washed and overlaid with 1 ml of DMEM-10 containing 1 µM [³H]GCV (12.4 Ci/mmol) in triplicate. Cells were incubated at 37°C for 30 h and harvested by trypsin dissociation, pelleting, and washing twice in serum-free DMEM. Replicate wells on the same plates were used for cell enumeration. All traces of supernatant were removed, and the pellets were extracted with ice-cold 14.3% perchloric acid. 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 high-performance liquid chromatography (HPLC) analysis.

Phosphorylation of ³H-nucleosides. Thirty micrograms of purified GST, GST-HHV8 TK, GST-EBV TK or GST-HSV-1 TK was used as the enzyme source in phosphorylation reactions as described above. The reaction mixtures contained either 29.8 µM [³H]dThd, 17.1 µM [³H]AZT, or 16.1 µM [³H]GCV as the substrate in a 100-µl final volume. Following a 60-min incubation at 37°C, the reactions were extracted with 220 µl 14.3% perchloric acid (Fisher) and neutralized with 15 µl of 1 M KHPO₄ and 25 µl of 4 M KOH. The precipitate was removed by centrifugation at 12,000 × g for 10 min, and 15 µl was used for HPLC analysis.

HPLC analysis. HPLC analysis was performed on a model 650E chromatograph with a model 600E system controller and a model 484 tunable absorbance detector (Waters). Nucleotides were separated on a POROS 10 SAX anion-exchange column (PerSeptive Biosystems, Framingham, Mass.), using a linear gradient of 5 mM NH₄H₂PO₄ buffer, 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 with 100% buffer B, 1 min of a linear gradient from 100% buffer B to 100% buffer A, and 2 min with buffer A, using 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 content of the vials were assayed for radioactivity by scintillation counting. To assign peaks and assess reproducibility, the retention time of unlabeled nucleosides was monitored by absorbance at 254 nm. Unlabeled AZT and GCV had identical retention times as [³H]AZT (15 to 30 s) and [³H]GCV (30 to 45 s), respectively. Samples containing [³H]AZT were spiked with 0.12 mM AZT-DP, which had the same retention time (3.2 to 4.0 min) as the peak assigned as [³H]AZT-DP. The retention times of GCV mono-, di-, and triphosphates were as reported previously (15). Picomoles of phosphorylated product was calculated by determining the counts per minute per picomole of each ³H-nucleoside before the assay. This procedure could detect 0.03 pmol of dThd.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

HHV8 ORF 21 has TK activity in mammalian cells. To determine if HHV8 ORF 21 encodes a protein with TK activity, pCMV-HHV8-TK was transfected into 143B TK cells and grown in HAT selection medium. pCMV- and pCMV-EBV-TK-transfected cells were included as negative and positive controls, respectively. The time to formation of distinct colonies varied. In the pCMV-EBV-TK-transfected cells, 535 distinct colonies were obtained following 1 week of HAT selection. In the pCMV-HHV8-TK-transfected cells, 23 small, distinct colonies were visible only after 2 weeks in selection medium. All vector control-transfected cells died following 1 week of selection. The HHV8 TK clones were thus 25-fold fewer in number than EBV TK clones and grew more slowly. Compared to the morphology of 143B TK cells cultured in medium without HAT and 143B EBV-TK cells selected in HAT medium, the 143B HHV8-TK cells selected in HAT medium were rounded up, grew in clumps, and did not reach confluence upon expansion. As the aminopterin component of HAT blocks the de novo synthesis of thymidylate (dTMP), cells growing in HAT medium must rely on the presence of a TK to provide a pool of dTMP to use in cellular DNA replication. The poor growth and condition of the HHV8 TK-transfected cells suggested that HHV8 TK might be an inefficient enzyme. We hypothesized that if the concentration of dThd in the medium (16 µM supplied in HAT supplement) is below the K_m of the enzyme, it may be inadequate to supply a pool of dTMP for cellular replication. The total concentration of dThd in the medium was therefore increased to 100 µM. This reduced the doubling time of the HHV8-TK-transfected cells, which allowed for their expansion in sufficient numbers to be used in subsequently described experiments. The cells were then routinely cultured in medium containing 100 µM dThd. The increased levels of dThd in the medium could not account for survival in HAT, as 143B TK cells grown in HAT medium containing 100 µM dThd could not survive (data not shown).

View larger version (47K): [in this window] [in a new window]   FIG. 1.   Detection of HHV8 TK transcripts by RNA blot hybridization. (Top) A 20-mer oligonucleotide probe antisense to ORF 21 was used to detect transcripts in total RNA (15 µg) from 143B cell clones transfected with HHV8 ORF 21 and selected in HAT-containing medium (lanes 1 and 2), 143B TK cells (lane 3), BCBL-1 cells (lane 4), and BCBL-1 cells treated for 72 h with NaB (3 mM) to induce the lytic cycle (lane 5). Position of the 28S and 18S rRNAs are indicated. The 2.5- and 4.4-kb HHV8 TK transcripts are indicated by arrows. (Bottom) The identical blot probed for human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control.

View larger version (27K): [in this window] [in a new window]   FIG. 2.   HHV8 TK mediates incorporation of [3H]dThd into high-molecular-weight DNA. 143B TK, 143B EBV-TK, and 143B HHV8-TK cells were incubated with 1 µM [3H]dThd. At 24 (), 48 (), and 72 () h, cells were extracted with TCA and acid-insoluble material was precipitated onto glass fiber filters. The total amount of radioactivity was determined by scintillation counting. Results represent the mean ± SD of three determinations.

HHV8 TK phosphorylates dThd less efficiently than EBV TK and HSV-1 TK. To examine the activity of HHV8 TK free from cellular contaminants, we expressed the protein in bacteria as a GST fusion protein and purified it by affinity chromatography. We obtained a protein of ~97 kDa as measured by SDS-PAGE analysis and Coomassie blue staining. This protein was judged to be >90% pure based on staining intensity (Fig. 3). An initial phosphorylation assay required >80 µg of GST-HHV8 TK protein to detect TK activity. Following buffer optimization as described in Materials and Methods, GST-HHV8 TK activity was routinely detected in assays using 10 µg of protein. GST alone, purified in the same manner as the fusion protein, had no TK activity, indicating that the purification method yielded protein free of bacterial TK (Fig. 4).

View larger version (42K): [in this window] [in a new window]   FIG. 3.   Expression and purification of HHV8 TK. GST-HHV8 TK and GST proteins were expressed in bacteria and purified by affinity chromatography on glutathione-Sepharose as described in the text. Thirty micrograms of each protein was analyzed on an SDS-10% polyacrylamide gel and visualized by Coomassie blue staining. Sizes are indicated in kilodaltons on the left.

View larger version (12K): [in this window] [in a new window]   FIG. 4.   Phosphorylation of dThd by purified HHV8 TK. The capacity of GST () or GST-HHV8 TK () (10 µg of each) to phosphorylate [3H]dThd was determined by disc assay as described in the text. Results are the mean ± SD of three determinations.

Only dThd and dThd analogs compete for phosphorylation by HHV8 TK. To characterize the HHV8 TK substrate specificity, a competition assay that allows rapid screening of a large number of compounds was used. Cold nucleoside analogs in a 10-fold molar excess over dThd were added to phosphorylation reactions and analyzed by disc assay as described in Materials and Methods. In this assay, inhibition of dThd phosphorylation is an indication that the nucleoside may fit into the enzyme active site and act as an alternative substrate. Of the nucleosides tested, only dThd and dThd analogs are recognized by the enzyme. The nucleoside dAdo, dGuo and its analogs GCV, ACV, PCV, and BCV, and dCyd and its analogs FMAC, dFdC, and CDV are not able to compete with dThd for phosphorylation by this enzyme (Fig. 5A). Although these data cannot completely exclude a given analog as a substrate for the enzyme, a 10-fold molar excess of an alternate substrate would likely compete with dThd. Competition assays were also performed with HHV8 TK without the GST moiety, and the resulting inhibition profile was identical (data not shown).

View larger version (2619K): [in this window] [in a new window]   FIG. 5.   Inhibition of GST-HHV8 TK activity by nucleoside analogs measured by competition with dThd. (A) The indicated nucleosides were included in phosphorylation reactions in 10-fold molar excess over dThd. (B) Phosphorylation reactions were performed with dThd alone (No Drug) or in the presence of GCV in 10-fold (GCV-10), 100-fold (GCV-100), and 1,000-fold (GCV-1000) molar excess over dThd. BVDU in 10-fold molar excess over dThd (BVDU-10) is included for comparison. Results are reported as percent activity of the enzyme with [3H]dThd alone.

GCV is not toxic to HHV8 TK-expressing cells. Phosphorylation of GCV in cells results in a decrease in cell viability and/or cell death due to incorporation of GCV into cellular DNA (40). To test whether HHV8 TK could sensitize cells to killing by GCV, 143B TK and 143B HHV8-TK cells were grown in the presence of increasing concentrations of GCV for 5 days. Cell viability was determined by enumeration of cells in the presence of trypan blue. 143B HSV1-TK and 143B EBV-TK cells were examined in parallel for comparison. As shown in Fig. 6, only 143B HSV1-TK cells are very sensitive to GCV. Above 5 µM GCV, toxic effects become increasingly evident in the other cell lines, consistent with the low-level phosphorylation of GCV in these cells (15). However, 143B HHV8-TK and 143B EBV-TK cells were not significantly more sensitive to GCV than 143B TK-cells at any GCV concentration tested. In addition, 143B HHV8-TK cells have been grown in continuous culture with 5 µM GCV in our laboratory for >30 days with no apparent cytotoxic effects compared to identical cells grown without GCV (data not shown). Thus, expression of HHV8-TK in 143B TK cells does not sensitize these cells to GCV. To correlate these results with GCV metabolism, the total amount of phosphorylated GCV was measured by HPLC analysis after a 30-h incubation of these cells in medium containing 1 µM [³H]GCV. The mean amounts of phosphorylated GCV from triplicate determinations were 0.04 pmol/10⁶ 143B TK cells, 0.09 pmol/10⁶ 143B EBV-TK cells, 0.11 pmol/10⁶ 143B HHV8-TK cells, and 83.2 pmol/10⁶ 143B HSV1-TK cells.

View larger version (61K): [in this window] [in a new window]   FIG. 6.   Comparative cytotoxicity of GCV. 143B TK cells () and 143B cells expressing HHV8 TK (), EBV TK (), and HSV-1 TK () were cultured for 5 days in the presence of GCV at the indicated concentrations. Viability was determined by enumeration of live cells in the presence of trypan blue. The number of live cells at each GCV concentration is expressed as the percentage of the same cells cultured for 5 days without GCV. Results are the mean ± SD of three determinations.

Purified HHV8 TK does not phosphorylate [³H]GCV but diphosphorylates [³H]AZT and [³H]dThd. The competition assay predicts that GCV is a poor substrate for HHV8 TK but indicates that AZT is likely a better substrate than dThd. To test this directly, phosphorylation assays were set up as in the disc assay with 30 µg of enzyme and nucleoside concentrations of 29.8 µM for [³H]dThd, 17.1 µM for [³H]AZT, and 16.1 µM for [³H]GCV. Reactions proceeded for 1 h at 37°C, and products were analyzed by HPLC analysis. Results shown in Table 2 illustrate several differences between the enzymes. First, evident is the low TK activity of HHV8 TK. Under similar experimental conditions, the enzyme phosphorylated 12.7 pmol of dThd, compared to 61.1 pmol of dThd by EBV TK and 68.7 pmol of dThd by HSV-1 TK. Interestingly, 5.7 pmol (45%) of the dThd phosphorylated by HHV8 TK is dThd-diphosphate (dTDP). Thus, the enzyme is more similar to HSV-1 TK (52.7 pmol of dTDP, 77%) than to EBV TK (0.5 pmol of dTDP, 0.8%) in its thymidylate kinase activity. This monophosphate kinase activity of HHV8 TK is even more pronounced in the phosphorylation of AZT. Of the 38.4 pmol of AZT phosphorylated by HHV8 TK, 29.9 pmol (78%) was phosphorylated to AZT-DP. This differs markedly from both EBV TK and HSV-1 TK, in which case 32 pmol (100%) and 31.4 pmol (96%) of the AZT, respectively, were phosphorylated to AZT monophosphate (AZT-MP). In the case of GCV, only HSV-1 TK could phosphorylate GCV to GCV-MP. There was no detectable phosphorylation of GCV by either HHV8 TK, as predicted, or EBV TK, as demonstrated previously (15), indicating that GCV either is not or is a very poor substrate for either enzyme.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

HHV8 ORF 21 encodes a protein with TK activity, as evidenced by the ability of 143B TK cells expressing ORF 21 to grow in HAT medium and to incorporate [³H]dThd into high-molecular-weight DNA. Measured by these assays, HHV8 TK was less active than EBV TK. Kinetic analysis indicated that the enzyme's K_m for dThd is ~33.2 µM. For the enzyme to work at its maximal velocity, the dThd concentration should be at least 66 µM. Thus, the decreased doubling time of 143B HHV8-TK cells upon growth in HAT-containing medium supplemented to 100 µM dThd compared to growth in unsupplemented HAT-containing medium (16 µM dThd) is consistent with the results of the biochemical analysis. If HHV8 TK is to supply the virus with a pool of dTMP to be used in replication, the level of dThd in the cell would need to be quite high. Interestingly, HHV8 also encodes a predicted thymidylate synthase (TS), as do HVS and varicella-zoster virus (47). The predicted HHV8 TS is 80% similar with the HVS TS, itself 70% similar to the human enzyme (18, 41). If this enzyme synthesizes dTMP, it may compensate in HHV8-infected cells for a low TK activity.

It has been suggested that inhibition of viral lytic replication may aid in preventing HHV8-associated disease by (i) decreasing the number of virions available to infect susceptible target cells or by (ii) preventing synthesis of replicative proteins that directly contribute to disease pathogenesis. For these reasons, the virus-specified TK is of interest as a drug target. Its ability to preferentially phosphorylate nucleoside analog drugs that selectively inhibit virus replication or destroy infected cells continues to be investigated. In this regard, the competition assay described here reveals the relative ability of nucleosides in 10-fold excess to compete at the HHV8 TK active site with the normal substrate dThd and may indicate the likelihood of a nucleoside being an alternate substrate. Thymidine analogs were the only nucleosides recognized by the HHV8 TK in this assay. This argues that the enzyme is not a nucleoside kinase with broad substrate specificity, such as is the case with human alphaherpesvirus TKs. Rather, these data are consistent with those of three other gammaherpesviruses, EBV, HVS, and bovine herpesvirus 4, whose TK enzymes have all been found to have limited substrate specificity (15, 19, 23, 48). The combined low activity and narrow specificity of the HHV8 TK demonstrated here may make it difficult to use as a drug target for prophylactic therapy. Nucleosides with a lower relative affinity than dThd may be of limited use, as the concentration needed to compete with dThd would likely be prohibitively high for nucleoside analogs that already may have significant toxicities.

Despite its low TK activity, the HHV8 TK has monophosphate kinase activity, distinct from that of EBV TK and HSV-1 TK, that could be therapeutically utilized with AZT and possibly other nucleosides. AZT toxicity is associated with a buildup of AZT-MP in normal cells due to phosphorylation of AZT by cellular TK and a subsequent inhibition of the cellular thymidylate kinase by AZT-MP (24, 27). This prevents accumulation of the active molecule AZT triphosphate (AZT-TP). Theoretically, AZT could be used as a selective reagent in HHV8-infected cells by bypassing the AZT-MP block via diphosphorylation of AZT by HHV8 TK. Following conversion of AZT-DP to AZT-TP by cellular enzymes, the net effect would be higher levels of AZT-TP accumulating in infected cells at lower initial doses of drug. In support of this, we have observed that AZT-TP is the major metabolite of AZT in 143B HHV8-TK cells and in BCBL-1 cells (37) treated to induce the lytic cycle (unpublished data). AZT-TP may then inhibit replication of the virus by inhibiting viral DNA polymerase as speculated for EBV (26) or could act as a selective cytotoxic reagent in infected cells by inhibition of cellular DNA polymerases (34).

The nucleoside analog GCV has been an important and effective antiviral agent in controlling human cytomegalovirus (CMV) disease (10), and it is important to clarify whether GCV can be used as an effective agent to control HHV8 disease as well. Cannon et al. (4) recently reported that human TK⁺ cells transiently transfected with ORF 21 were sensitized to GCV, and they found manyfold-higher levels of phosphorylated GCV in HHV8 TK-expressing cells than control cells, despite finding no TK activity associated with the enzyme. In our experiments, cells expressing HHV8 TK were found to have <3-fold more phosphorylated GCV than control cells when cultured in the presence of 1 µM GCV. In sharp contrast, cells expressing HSV-1 TK had >2,000 fold more phosphorylated GCV than control cells and >750-fold more than cells expressing HHV8 TK. Unlike HSV-1 TK-expressing cells, HHV8 TK-expressing cells were not more sensitive than control cells to any concentration of GCV tested. The discrepancy between results may be due to differences in cell types, protein expression levels, or the amount of GCV used in the experiments. However, GCV was not a substrate for a highly purified GST-HHV8 TK and could not compete with dThd for phosphorylation by the same enzyme. We therefore conclude that at the most, GCV is a very poor substrate for HHV8 TK.

The slightly higher levels of phosphorylated GCV in HHV8 TK-expressing cells compared to control, despite no detectable phosphorylation by purified enzyme, could indicate that other factors are involved in activity of this enzyme. We note that the N-terminal regions of gammaherpesviruses contain a domain of several hundred amino acids which have no homology to other herpesvirus TKs (17). These domains are not closely related between EBV and HHV8, and computer search reveals no recognizable motifs (11, 44), although Pro, Gly, and Ser constitute 33% of the various domains. The possibility that this region of HHV8 TK can interact with a cellular protein (or viral protein in infected cells) that affects its activity or ability to recognize nucleosides cannot be discounted.

The inability of HHV8 TK to phosphorylate GCV efficiently is not incompatible with reports of GCV inhibiting viral replication in vitro (22, 29, 33) or its efficacy in preventing KS (28, 30) for several reasons. The HHV8 DNA polymerase may be very sensitive to GCV triphosphate (GCV-TP), and the low levels of GCV-TP produced in cells, either by viral or by cellular enzymes, may be enough to inhibit the polymerase and thus viral replication. Further work is necessary to determine the sensitivity of the viral DNA polymerase to GCV-TP and other nucleoside triphosphates. While results of retrospective studies of GCV have been mixed (9, 14, 20, 21, 30, 39), in a recent prospective analysis, AIDS patients administered high doses of GCV that effectively prevented CMV retinitis were unexpectedly found to develop significantly less KS (28). CMV infection is highly immunosuppressive, and active disease in patients who are immunosuppressed often results in development of additional opportunistic infections, including those caused by DNA tumor viruses (13, 16). Thus, control of KS may be an indirect result of CMV eradication. It is also possible that another viral gene product, such as ORF 36, the virus phosphotransferase homologue of human CMV UL97, can phosphorylate GCV and be responsible for its activity in vivo (4). Alternatively, because of GCV superior toxicity, preferential uptake into rapidly growing tumor cells and conversion by cellular enzymes may create a situation where GCV acts as a chemotherapeutic reagent, similar to other nucleoside analogs, such as cytosine arabinofuranoside and adenine arabinofuranoside, that have both antiviral and chemotherapeutic activity (1, 35, 37, 50).

In summary, the HHV8 ORF 21 encodes a TK with low enzymatic activity. Its reactivity in competition assays indicates it has limited substrate specificity similar to other gammaherpesvirus TKs. The inability of the HHV8 TK to efficiently phosphorylate GCV further demonstrates the enzyme's limited specificity but does not necessarily preclude the usefulness of this drug in treating KS. The monophosphate kinase activity associated with HHV8 TK may present an exploitable property of the enzyme for future drug development. Characterization of the effects of nucleoside antivirals on the distinct cell populations productively infected or immortalized by HHV8 will permit optimization of therapeutic strategies that may prove distinct for latently infected cells.