Primary effusion lymphoma (PEL) is a rare B-cell lymphoma caused by Kaposi's sarcoma-associated herpesvirus (KSHV). PEL is poorly responsive to standard cytotoxic chemotherapy and portends a poor survival. Consequently, new effective treatment options are urgently needed. It is known that KSHV encodes two lytic genes, ORF36 (phosphotransferase) and KSHV ORF21 (thymidine kinase), which can phosphorylate ganciclovir and azidothymidine, respectively. Here, we have explored whether these genes can be used as therapeutic targets for PEL. PEL arises in pleural spaces and other effusions that provide a hypoxic environment. Based on Northern blot analysis, exposure of PEL cells to hypoxia up-regulated the expression of both ORF36 and ORF21. Using a newly developed nonradioactive reverse-phase high-performance liquid chromatography/mass spectrometry method to separate and quantify the phosphorylated forms of ganciclovir and azidothymidine, we found that PEL cells exposed to hypoxia produced increased amounts of the toxic triphosphates of these drugs. Moreover, we found that hypoxia increased the cell toxicity of ganciclovir and azidothymidine in PEL cells but had no significant effect on the herpesvirus-negative cell line CA46. These findings may have clinical applicability in the development of effective therapies for PEL or other KSHV-related malignancies. [Cancer Res 2007;67(14):7003–10]
- Human herpesvirus 8
Primary effusion lymphoma (PEL) is a rare B-cell lymphoma that develops particularly in patients with human immunodeficiency virus infection ( 1, 2). PEL characteristically arises in the pleural space and other body cavities and is sometimes referred to as body cavity lymphoma. Essentially all PEL tumors are infected with Kaposi's sarcoma-associated herpesvirus (KSHV), also called human herpesvirus-8, and most cases of PEL are coinfected with EBV ( 1– 3). KSHV plays a critical role in the pathogenesis of PEL ( 3, 4), as well as Kaposi's sarcoma and multicentric Castleman's disease ( 3, 5, 6). Although the incidence of these diseases has decreased since the introduction of highly active antiretroviral therapy in the developed world, they have not been eliminated. Moreover, KSHV-related malignancies are a major cause of morbidity and mortality in Africa and other parts of the developing world ( 7). There is currently no standard therapy for PEL. Although some cases respond to multiagent cytotoxic chemotherapy, the tumor is often resistant to such approaches and most patients die within months ( 8– 10). Consequently, new therapies for PEL are urgently needed.
In viral-induced tumors, viral genes are unique to the tumor cells and thus offer potential targets for attack ( 11). Because most herpesvirus-induced tumors have latently infected cells, such approaches would generally require targeting latent genes or inducing lytic replication. Although KSHV is in a latent state in most cell lines derived from PEL tumors, some KSHV lytic genes are activated in tumor cells in the body ( 12, 13) and there is evidence that certain lytic genes play a key role in the pathogenesis of PEL and other tumors ( 4, 14). Our group has found previously that KSHV is activated to lytic replication by hypoxia ( 15); in addition, we and others showed that certain lytic genes are specifically activated by hypoxia-inducible factor (HIF)–responsive elements (HRE) in their promoter regions ( 15– 18). Pleural and other body cavity effusions are relatively hypoxic environments ( 19, 20), and this may contribute to the lytic gene activation found in PEL tumor cells. One gene that is activated directly by hypoxia is ORF36 ( 16, 17). ORF36 has kinase activity ( 21), may be involved in cell signaling, ( 22, 23), and has been reported to activate ganciclovir to a toxic triphosphate moiety ( 24). Another KSHV lytic gene, ORF21, encodes a thymidine kinase that can phosphorylate zidovudine (azidothymidine) and to a variable extent ganciclovir ( 24, 25). These observations suggested that these lytic genes might be used as specific targets for PEL therapy.
Here, we report on the activation of these KSHV genes in PEL lines by hypoxia, their phosphorylation of ganciclovir and azidothymidine, and the enhanced phosphorylation of these drugs in PEL lines exposed to hypoxia. In addition, we explore the potential of using this type of targeting to develop a novel therapy for PEL.
Materials and Methods
Drugs and other reagents. Ganciclovir from Roche Laboratories was prepared (20 mmol/L) in PBS. Azidothymidine from Sigma and ganciclovir triphosphate (GCVTP) from TriLink Biotechnologies were prepared (10 mmol/L) in PBS and stored at −20°C. Azidothymidine triphosphate (AZTTP) from Calbiochem was prepared (10 mmol/L) in PBS and stored at −70°C. Venom phosphodiesterase was obtained from Worthington Biochemical Corp. Acetate kinase was obtained from Roche Applied Sciences.
Synthesis of ganciclovir monophosphate, ganciclovir diphosphate, azidothymidine monophosphate, and azidothymidine diphosphate. Venom phosphodiesterase was used to synthesize ganciclovir monophosphate (GCVMP) from GCVTP ( 26). The reaction mixture consisted of 0.01 mol/L Tris-HCl, 0.01 mol/L MgCl2 (pH 9.0), 1 unit venom phosphodiesterase, and 250 μmol/L GCVTP. The reaction proceeded for 6 h at 37°C and was inactivated by heating for 2 min at 95°C. Generation of GCVMP was confirmed by liquid chromatography/mass spectrometry (LC/MS). Azidothymidine monophosphate (AZTMP) was synthesized using the same method. Acetate kinase was used to specifically synthesize ganciclovir diphosphate (GCVDP) from GCVTP ( 26). The reaction mixture consisted of 0.01 mol/L Tris-HCl, 0.01 mol/L MgCl2, 0.2 mol/L ammonium acetate (pH 8.0), 500 μg/mL acetate kinase, and 250 μmol/L GCVTP. The reaction proceeded for 6 h at 37°C and was heat inactivated at 95°C for 2 min. Generation of GCVDP was confirmed by LC/MS. Azidothymidine diphosphate (AZTDP) was synthesized using the same method as that for GCVDP.
Separation of phosphorylated forms of azidothymidine and ganciclovir by reverse-phase high-performance LC. Ganciclovir nucleotides were separated on a Develosil RP-Aqueous 5-μm column (250 mm × 4.6 mm) from Phenomenex. The mobile phase was a 25 mmol/L ammonium formate (pH 9.0) buffer/acetonitrile mixture. Ammonium formate (pH 9.0) buffer was run from 0 to 7 min followed by a gradient from 0% to 70% acetonitrile from 7 to 12 min and then another gradient from 70% to 95% acetonitrile from 12 to 24 min followed by reequilibration to 0% acetonitrile for the remaining time. GCVMP, GCVDP, and GCVTP were eluted at 14, 11.5, and 10 min, respectively. A slightly modified method was used to separate azidothymidine phosphorylated metabolites. The starting buffer was a mixture of 98% ammonium formate buffer and 2% acetonitrile (percentage acetonitrile optimized for each column using known standards) from 0 to 7 min followed by the same gradients described previously. AZTMP, AZTDP, and AZTTP were eluted at 19.5, 17, and 11 min, respectively. The AZTTP retention time shifted to 18 min in the presence of magnesium ions; therefore, the concentrations of both forms were included when analyzing cell extracts.
Cell lines. Except when indicated otherwise, BCBL-1 cells (from the NIAID AIDS Research and Reagent Program, Rockville, MD) and JSC-1 cells (gift of Dr. Richard Ambinder, Johns Hopkins University, Baltimore, MD) were cultured in RPMI 1640 (Life Technologies) with 10% heat-inactivated FCS (Hyclone), 100 units/mL penicillin, 100 μg/mL streptomycin sulfate, and 2 mmol/L l-glutamine (other cell culture materials were from Life Technologies). CA46 and 293T cells (American Type Culture Collection) were maintained in DMEM (Life Technologies) with 10% heat-inactivated FCS, 100 units/mL penicillin, and 100 μg/mL streptomycin sulfate.
Determination of ganciclovir/azidothymidine cytotoxicity and synergy. JSC-1 and BCBL-1 cells were seeded at 100,000 live cells/mL in 96-well plates (200 μL/well) and incubated for 24 h in normoxia. Cells were treated with PBS, ganciclovir, or azidothymidine or with a combination of the two drugs. Cells were then incubated for an additional 72 h in either normoxia (21% O2) or hypoxia (1% O2). After 72 h, cell viability was assessed using the Cell-Glo Luminescent Cell Viability Assay (Promega). Luminescence was read by the Victor2 luminometer from Perkin-Elmer. To determine if the combination of ganciclovir and azidothymidine resulted in a synergistic toxic effect, dose-response curves were run using various doses of ganciclovir and azidothymidine (10 μmol/L). Results were analyzed using the CalcuSyn program (Biosoft).
Plasmid constructs and cell transfection with ORF21 and ORF36. The coding sequences of ORF21 and ORF36 in tandem with the coding sequence for green fluorescent protein (GFP) were PCR amplified from BCBL-1 cellular DNA with primers containing Bgl11 and EcoR1 restriction sites at the 5′ and 3′ ends, respectively. ORF21 was amplified with the primer pairs 5′-CAGagatctATGGCAGAAGGCGGTTTT-3′ (ORF21F) and 5′-ATAgaattcGACCCTGCATGTCTCCTC-3′ (ORF21R) and ORF36 was amplified with the primer pairs 5′-CATagatctATGCGCTGGAAGAGAATG-3′ (ORF36F) and 5′-ATAgaattcGAAAACAAGTCCGCGGGT-3′ (ORF36R). The lower case letters indicate Bgl11 and EcoR1 restriction sites, respectively. Amplified products were digested with Bgl11 and EcoR1 and cloned into the same restriction sites of the expression vector pEGFP-C1 (Clontech) to construct pEGFP-36 and pEGFP-21. For transfection, 293T cells were trypsinized, centrifuged, and resuspended in fresh medium. They were plated at 200,000 live cells/mL (2 mL/well) in 12-well plates and incubated overnight at 37°C in normoxia. For each well, a transfection mixture was made containing 97 μL serum-free DMEM and 3 μL FuGene 6 transfection reagent from Roche. This mixture was incubated at room temperature for 5 min and 1 μg plasmid DNA encoding ORF21 or ORF36 was added. The mixture was again incubated at room temperature for 15 min and then added dropwise to the appropriate culture well. The cells were again incubated for 24 h and treated with PBS, azidothymidine, ganciclovir, or azidothymidine plus ganciclovir. The cells were cultured for 48 h and removed from the bottom of the plate by vigorous mixing; cells and supernatant were transferred to a microfuge tube and extracted as described.
Analysis of ganciclovir or azidothymidine phosphorylation in PEL cells and uninfected B-cells. BCBL-1, JSC-1, and CA46 cells were prepared at 200,000 live cells/mL and placed in 25-cm2 flasks (10 mL/flask). Cells were incubated at 37°C in normoxia for 24 h before treatment with azidothymidine and/or ganciclovir and then incubated for another 48 h with drugs or control medium in either normoxia or hypoxia. Cells were pelleted by centrifugation at 1,500 × g for 10 min and then washed twice with PBS. Cell pellets were resuspended in 300 μL of 60% methanol and heated for 2 min at 95°. The extracts were centrifuged at 10,000 rpm for 10 min and the supernatants were transferred to a new tube and dried in a Speedvac Plus from Thermo. The residue was resuspended in 25 μL of 25 mmol/L ammonium formate (pH 9.0) immediately before injection, and 20 μL were injected onto the column. This method was also used on the transfected 293T cells treated with azidothymidine and/or ganciclovir.
RNA isolation and Northern blot analysis. Total cellular RNA was isolated from PEL cells cultured for 24 h under various conditions by using Trizol reagent (Invitrogen). Northern blot hybridization was done using a nonisotopic digoxigenin labeling probe generated by PCR (Roche Applied Sciences). Briefly, 5 μg of total RNA were fractionated on 1% agarose gel containing 2% formaldehyde and subsequently transferred to Hybond-N nylon membranes (Amersham) in the presence of 20× SSC overnight by capillary transfer. After fixing the RNA to the membrane by UV cross-linking, the membranes were prehybridized for 30 min and hybridized with DIG Easy Hyb buffer (Roche Applied Science) overnight at 50°C with full-length digoxigenin-labeled gene-specific DNA probes incorporated by PCR. The membranes were washed, incubated for 30 min with an appropriate dilution of anti-digoxigenin antibody conjugated with alkaline phosphatase, washed again, incubated with CDP-Star solution (Roche Applied Sciences) for 5 min, and exposed to film with an intensifying screen. Membranes were stripped (50% formamide, 2× saline-sodium phosphate-EDTA) at 65°C for 1 h and then rehybridized with β-actin probe generated from KSHV cDNA as a loading control.
Analysis of KSHV by Western blot. JSC-1 cells were plated at 250,000 cells per mL, incubated overnight, treated with 12-O-tetradecanoylphorbol-13-acetate (TPA; 5 nmol/L), TPA plus ganciclovir (100 μmol/L), or TPA plus ganciclovir (100 μmol/L) and azidothymidine (50 μmol/L), and incubated in normoxia or hypoxia for 48 h. Cells were centrifuged (10 min at 1,500 × g at 4°C) and the resulting supernatant was ultracentrifuged at 75,000 rpm for 45 min at 4°C. The pellet was washed with 5 mL PBS and pelleted again by ultracentrifugation. Pellets were resuspended in lithium dodecyl sulfate sample buffer (Novex) and boiled at 95°C for 2 min. Equal volumes of the resuspended pellet were electrophoresed on a 4% to 12% bis-Tris polyacrylamide gel with MOPS buffer by means of the Nupage system (Novex). Proteins were electroblotted onto nitrocellulose membrane and blocked in 5% dried milk/TBS with 0.2% Tween 20 for 1 h at room temperature. After blocking, the membrane was probed with a monoclonal antibody (mAb) to ORF45 (gift of Dr. Yan Yuan, University of Pennsylvania, Philadelphia, PA) for 2 h at room temperature, washed, and incubated for 30 min with an anti-mouse secondary antibody conjugated to alkaline phosphatase. Bands were detected with Western Blue–stabilized alkaline phosphatase substrate (Promega).
Assessment of interactions between ganciclovir and azidothymidine. To determine the potential interaction of azidothymidine and ganciclovir in PEL cells, toxicity dose response curves were assessed for each drug alone to obtain the median effect dose (Dm) and the sigmoidicity of the dose-effect curve (m). In addition, the effect of ganciclovir in combination with azidothymidine was assessed. Analysis of the data was done using CalcuSyn (Biosoft) to determine the combination index (CI), a measure of synergy or antagonism between two drugs. A CI of 1 indicates no synergy, a CI of >1 indicates antagonism, and a CI of <1 indicates synergy.
Activation of KSHV ORF21 and ORF36 by hypoxia. We first explored the up-regulation by hypoxia of ORF21 and ORF36 ( 24, 25). Our group showed previously that KSHV is activated to lytic replication by hypoxia ( 15) and that ORF36 is in a cluster of genes that is specifically up-regulated by hypoxia through an HRE in the promoter region ( 16, 17). As shown in Fig. 1 , Northern blot analysis revealed that both ORF21 and ORF36 mRNA were up-regulated when JSC-1 cells were exposed to hypoxia. ORF36 was detectable within 8 h after exposure to hypoxia but was more strongly induced at 24 h and reached levels similar to those obtained with the chemical inducer butyrate. ORF21 mRNA was undetectable 8 h after exposure to either hypoxia or butyrate. However, ORF21 mRNA was detectable 24 h after exposure to either hypoxia or butyrate at approximately equal levels, although the bands were fainter than those for ORF36 ( Fig. 1). Thus, ORF21 and ORF36 are both up-regulated when cells are exposed to hypoxia, although the induction of ORF21 is less robust and somewhat delayed.
Phosphorylation of azidothymidine and ganciclovir by ORF21- and ORF36-transfected cells. To assess the ability of cells expressing ORF21 and/or ORF36 gene products to phosphorylate azidothymidine or ganciclovir, we first developed a nonradioactive reverse-phase high-performance LC/MS (RP-HPLC/MS) method for separating and identifying their phosphorylated products. Using this technique, the phosphorylated forms of ganciclovir and azidothymidine could be separated and identified based on elution time coupled with detection of target masses by selective ion monitoring ( Fig. 2A and B ). Dose response curves (1–100 pmol AZTTP) confirmed a strong linear correlation between the concentration of nucleotides injected and the mass signal obtained (linear correlation coefficient 0.999 for 0–100 pmol AZTTP and 0.998 for 0–100 pmol GCVTP).
RP-HPLC/MS was used to assess the ability of cells transfected with ORF21 or ORF36 to phosphorylate azidothymidine and ganciclovir. Cells (293T) were transfected with plasmids containing GFP-tagged ORF21, GFP-tagged ORF36, or a vector control and incubated for 48 h with 25 μmol/L azidothymidine or 500 μmol/L ganciclovir. Cells were extracted and analyzed by RP-HPLC/MS. For cells exposed to ganciclovir, the most abundant phosphorylated form was, in all cases, GCVTP ( Fig. 2C). ORF21- and ORF36-transfected cells were both capable of producing greater levels of all the metabolites of ganciclovir than the vector-transfected cells ( Fig. 2C). Interestingly, ORF36-transfected cells generated 2.5-fold more GCVTP (76 pmol/million cells) than the ORF21-transfected cells, although they produced similar levels of the monophosphate and diphosphate ( Fig. 2C).
Consistent with previous studies on uninfected cells treated with azidothymidine ( 27), the most abundant phosphorylated form of azidothymidine produced in 293T cells transfected with a control vector plasmid and treated with 25 μmol/L azidothymidine was AZTMP ( Fig. 2D). However, when cells were transfected with ORF21, the concentrations of AZTDP and AZTTP increased substantially and accumulated to substantially higher levels than the monophosphate ( Fig. 2D). The most abundant form of azidothymidine identified in ORF21-transfected 293T cells was AZTTP. These cells produced 437 pmol/million cells of AZTTP compared with a much lower level for vector-transfected cells (2.9 pmol/million cells; Fig. 2D). Interestingly, ORF36-transfected cells did not produce greater amounts of either AZTDP or AZTTP than the vector-treated cells, indicating that azidothymidine is a poor substrate for this enzyme under these conditions ( Fig. 2D). These results indicate that KSHV ORF21 can phosphorylate azidothymidine and to a lesser extent ganciclovir, whereas ORF36 is only able to phosphorylate ganciclovir efficiently.
Phosphorylation of azidothymidine and ganciclovir in PEL cells exposed to hypoxia. As noted above, hypoxia up-regulates the expression of ORF21 and ORF36. With this background, we hypothesized that the phosphorylation of ganciclovir and azidothymidine would be increased when PEL cells were exposed to hypoxia. To test this hypothesis, we exposed JSC-1 cells to TPA or hypoxia, treated them with ganciclovir or azidothymidine for 48 h, and analyzed cell extracts by RP-HPLC/MS. KSHV-uninfected CA46 cells were used as a control. In normoxia, JSC-1 cells produced low levels of GCVDP and GCVTP (5 and 18 pmol/million cells, respectively) and no detectable GCVMP (<1 pmol/million cells; Fig. 3A ). However, when JSC-1 cells were exposed to hypoxia, GCVDP and GCVTP increased by ∼3-fold to 13.4 and 51.2 pmol/million cells, respectively. In addition, GCVMP was now detectable at a level of 7.7 pmol/million cells. These increases were comparable with those observed when the cells were exposed to 10 nmol/L TPA (2.7-fold increase in GCVDP and 3.9-fold increase in GCVTP; Fig. 3A). By contrast, in CA46 cells, all three metabolites of ganciclovir were detected at low levels but there were no significant increases following exposure to TPA or hypoxia ( Fig. 3A).
When CA46 or JSC-1 cells were treated with 25 μmol/L azidothymidine, the most abundant phosphorylated form detected was, as expected, AZTMP ( Fig. 3B). Interestingly, whereas CA46 cells produced very low levels of the toxic triphosphate, AZTTP, (<1 pmol/million cells under all conditions), even normoxic JSC-1 cells produced measurable quantities of AZTTP (7 pmol/million cells; Fig. 3B). The levels of AZTTP further increased when JSC-1 cells were exposed to hypoxia (12 pmol/million cells) or TPA (21 pmol/million cells; Fig. 3B). Thus, exposure of JSC-1 cells to hypoxia resulted in increased production of both GCVTP and AZTTP compared with cells cultured in normoxia or CA46 cells exposed to hypoxia.
Ganciclovir and azidothymidine are toxic to PEL cells and this effect is enhanced when the cells are exposed to hypoxia. The cytotoxic effects of azidothymidine, ganciclovir, and their combination were examined in PEL cell lines incubated in normoxia or hypoxia. Cells were exposed to various concentrations of the drugs for 72 h and percentage viability was assessed by the ATP luminescence assay. As shown in Fig. 4A , the 50% cytotoxic concentration (CC50) of ganciclovir in JSC-1 cells was ∼250 μmol/L but the CC50 decreased to ∼150 μmol/L when cells were incubated in hypoxia. For BCBL-1 cells, the CC50 of ganciclovir in normoxia was ∼430 μmol/L ( Fig. 4B). Again, in hypoxia, the toxicity of ganciclovir was increased at each concentration tested, and the CC50 decreased to ∼300 μmol/L. By contrast, ganciclovir had relatively little toxicity in CA46 cells at concentrations of up to 500 μmol/L in either normoxia or hypoxia ( Fig. 4C).
Because the phosphorylation of both ganciclovir and azidothymidine were increased in KSHV-infected cells by hypoxia, we examined their interaction in inducing toxicity. Azidothymidine alone (10 μmol/L) induced little or no toxicity in PEL cells in normoxia and also induced only a low level of toxicity in hypoxia (20% killing of JSC-1 cells and 5% of BCBL-1 cells). However, 10 μmol/L azidothymidine did increase the toxicity of ganciclovir in JSC-1 cells shifting the CC50 from ∼250 μmol/L ganciclovir to 150 μmol/L ganciclovir in the presence of azidothymidine ( Fig. 5A ). Moreover, when the JSC-1 cells were cultured under conditions of hypoxia, 10 μmol/L azidothymidine increased the toxicity further with a CC50 of 50 μmol/L for ganciclovir ( Fig. 4A). Again, similar results were seen in BCBL-1 cells. In normoxia, the addition of 10 μmol/L azidothymidine decreased the CC50 of ganciclovir from ∼430 to ∼350 μmol/L. In addition, in the presence of 10 μmol/L azidothymidine, exposure of BCBL-1 cells to hypoxia substantially increased the toxicity of ganciclovir with a CC50 of ∼175 μmol/L ( Fig. 4B). By contrast to the results with these PEL cell lines, the combination of 10 μmol/L azidothymidine and up to 500 μmol/L ganciclovir induced little toxicity in CA46 cells when cultured in either normoxia or hypoxia ( Fig. 4C). Analysis of JSC-1 cells by Annexin V staining as described previously ( 28) showed that the increased killing by ganciclovir in hypoxia was due to apoptotic cell death. After 48 h in hypoxia, 33% of the cells were apoptotic and this increased to 49% with exposure to 100 μmol/L ganciclovir (data not shown).
To assess whether ganciclovir and azidothymidine were synergistic, antagonistic, or additive in their toxicity to PEL cells, further experiments were done and the results were analyzed using CalcuSyn to determine the extent of synergy. In normoxia, synergy was only observed at the two highest drug concentrations when ganciclovir and azidothymidine were tested at a ratio of 10:1. ( Table 1 ). However, in hypoxia, the two drugs were synergistic at all concentrations tested with the CI ranging from 0.29 to 0.74, consistent with the hypoxic activation of increased phosphorylation of these drugs. Synergy between the two drugs was also observed in JSC-1 cells in normoxia and hypoxia with all but the lowest dose of ganciclovir (25 μmol/L) when a constant dose of azidothymidine (10 μmol/L) was used (CI range, 0.1–0.7; data not shown). Overall, the results indicate that cells infected with KSHV are more sensitive to ganciclovir and azidothymidine than uninfected cells, that hypoxia can further sensitize the infected cells to killing by these drugs, and that the two drugs are synergistic in their effects under conditions of hypoxia.
Ganciclovir and ganciclovir/azidothymidine inhibit KSHV production from JSC-1 cells. Lytic activation of PEL cells leads to virus production and eventual cell lysis ( 29). To determine if ganciclovir or ganciclovir/azidothymidine treatment resulted in the inhibition of virus production, we analyzed the production of virus in the supernatant of JSC-1 cells following treatment with TPA or hypoxia in the presence or absence of ganciclovir (100 μmol/L) or ganciclovir (100 μmol/L) plus azidothymidine (50 μmol/L). Virus in pelleted supernatant was assessed using a mAb to ORF45 protein, a protein contained within KSHV particles ( 30). In the absence of TPA or hypoxia, only a very low level of virus production from JSC-1 cells was detected, and this was inhibited by ganciclovir ( Fig. 5, lanes 1 and 2). Exposure of cells to hypoxia led to an increase in virus production that was inhibited by ganciclovir or ganciclovir/azidothymidine ( Fig. 5, lanes 3–5). Exposure of cells to TPA in normoxia led to strong induction of virus that was again substantially inhibited by ganciclovir or ganciclovir/azidothymidine ( Fig. 5, lanes 6–8). Thus, under conditions of hypoxia, ganciclovir and ganciclovir/azidothymidine could simultaneously inhibit KSHV virus production while still resulting in substantial toxicity to PEL cells.
It was shown previously that transfection of KSHV ORF21 or ORF36 into cells could result in increased phosphorylation of azidothymidine or ganciclovir, respectively ( 24, 25). In this report, we confirmed these findings using a novel nonradioactive RP-HPLC/MS method and explored the possibility of using these viral genes to specifically kill PEL cells. We showed that expression of ORF21 and ORF36 mRNA was increased in PEL cells exposed to hypoxia, that such cells have increased production of the toxic moieties of azidothymidine and ganciclovir, and that exposure of PEL cells to azidothymidine and ganciclovir in the presence of hypoxia resulted in increased killing of cells.
The RP-HPLC/MS methodology used here enabled us to simultaneously examine the monophosphate, diphosphate, and triphosphate products of azidothymidine and ganciclovir. In control 293T cells exposed to azidothymidine, the most abundant form was AZTMP. This is consistent with previous reports of azidothymidine metabolism and occurs because AZTMP is a poor substrate for thymidylate kinase, the cellular enzyme responsible for phosphorylation to the diphosphate ( 27). By contrast, transfection of ORF21 changed the relative production of the monophosphate and triphosphate moieties, resulting in a relative increase in AZTTP. ORF21 can catalyze the formation of both the monophosphate and the diphosphate of azidothymidine ( 25), suggesting that the increased levels of AZTTP observed results from a direct production of AZTDP by ORF21 and subsequent phosphorylation to AZTTP by cellular enzymes. Although AZTTP has a degree of specificity for reverse transcriptase, it can also inhibit DNA polymerase, in particular DNA polymerase γ ( 31, 32), and this indicated that azidothymidine might selectively kill or suppress the growth of KSHV-infected cells exposed to hypoxia. GCVTP is also toxic and can kill tumor cells ( 33, 34). It is worth pointing out that in the current study, we found that ganciclovir is phosphorylated by ORF21 in addition to being phosphorylated by ORF36. This is somewhat at variance with the finding of Gustafson et al. ( 25) using purified enzyme; however, it is consistent with the analysis of Cannon et al. ( 24) using tritiated ganciclovir in transfected cells and may indicate that certain cellular factors play a role in supporting the ORF21 phosphorylation of ganciclovir.
In this study, the ganciclovir CC50 in BCBL-1 cells was consistent with previous reports ( 35, 36). Although a previous study did not show ganciclovir toxicity when BCBL-1 cells were activated with TPA ( 36), we found that the toxicity of ganciclovir was enhanced by hypoxia or with the addition of 10 μmol/L azidothymidine, thereby shifting the CC50 toward the IC50 for KSHV ( 35). Half of the JSC-1 cells in hypoxia were killed in the presence of 50 μmol/L ganciclovir and 10 μmol/L azidothymidine. Moreover, under most, but not all, conditions tested, there was synergy between ganciclovir and azidothymidine. Although it was possible that the synergy observed with azidothymidine was due to an induction of KSHV gene expression by azidothymidine like that reported with EBV ( 37), the concentrations of azidothymidine we used did not induce RTA in PEL cells (data not shown).
There has been an interest in using vectors encoding herpes simplex thymidine kinase in combination with ganciclovir or other nucleoside therapy as experimental tumor therapy ( 33, 34). Several groups have shown that KSHV replication is sensitive to treatment with nucleoside analogues ( 38– 40). As essentially all PEL cells are infected with KSHV, viral thymidine kinase (ORF21) and phosphotransferase (ORF36) potentially represent built-in gene targets for selective therapy. ORF36 is particularly of interest, as it can be directly induced by hypoxia through a HRE in the promoter region of the ORF34 to ORF37 viral gene cluster, potentially bypassing the need for lytic activation through the Rta lytic switch ( 17). Because PEL cells grow in a hypoxic environment, the potential for direct ORF36 expression exists. The experiments here provide evidence to support the use of azidothymidine and ganciclovir in the setting of hypoxia. They also suggest that azidothymidine and ganciclovir might be useful in combination to treat MCD, as a high percentage of these tumor cells have KSHV lytic gene activation ( 12, 13).
Two main concerns arise when considering the use of these antiviral drugs for the treatment of PEL or MCD. First, can a high enough level of these drugs be obtained in patients and, second, would the extent of killing obtained then be sufficient? The effective concentrations of ganciclovir (50–150 μmol/L) and azidothymidine (10–50 μmol/L) in this study are substantially higher than those attained with the usual clinical doses of these drugs. However, during short-term dosing, ganciclovir serum levels of 50 μmol/L or higher can be attained ( 41). In addition, although serum levels of ∼2.5 to 3.5 μmol/L azidothymidine are attained with the usual p.o. dose of 250 mg ( 42), substantially higher serum levels (9–15 μmol/L) can be attained and tolerated over a short period ( 42, 43). Additionally, it may be possible to attain locally higher levels of the drugs through direct delivery to the body cavities involved with PEL.
On the issue of the degree of killing, the data here suggest that the drugs will preferentially kill those cells in which there is either complete lytic activation or activation of ORF36 (or ORF21), as occurs in the presence of hypoxia. PEL tumors grow in a hypoxia environment ( 19, 20), and as a result, there can be direct activation of ORF36. Although only a small percentage of PEL tumor cells may be undergoing lytic activation, there is evidence of expression of certain lytic genes, such as viral IL-6, by a relatively greater percentage of these cells ( 12, 13). Moreover, there is evidence that certain lytic genes of KSHV, such as viral IL-6, or cellular factors that are up-regulated by KSHV lytic genes promote PEL pathogenesis through a paracrine mechanism ( 4, 14). Thus, a substantial antitumor effect may possibly be gained by selectively killing the PEL cells that express these lytic genes. The finding by Ghosh et al. ( 44) of a long-lasting remission in a patient with PEL induced by the combination of twice-daily parenteral azidothymidine (1.5 gm/dose) and daily dosing with 5 million units IFN-α supports the potential clinical utility of this approach.
The use of azidothymidine and ganciclovir in PEL could be combined with other approaches ( 45) that can enhance KSHV kinase expression or kill the cells through other mechanisms ( 44, 46). It may also be possible to increase HIF in PEL cells through the use of hypoxia-mimics like desferrioxamine ( 15). In addition, various agents, such as bortezomib and valproic acid, have been shown to activate the KSHV lytic cycle. For bortezomib, cell killing was enhanced by coculture with ganciclovir ( 47, 48). Finally, the feasibility of the approach described here is bolstered by reports that patients with EBV-associated Burkitt lymphoma or central nervous system lymphoma can be effectively treated with azidothymidine-based therapy ( 37, 49, 50).
Viruses that cause tumors may provide specific virus-encoded targets that can be used to develop specific therapies. In this article, we have provided evidence that activation of KSHV-encoded kinases by hypoxia might be used as a means to specifically target PEL cells by drugs activated by the kinases. Such an approach may be worth exploring in this usually fatal disease.
Grant support: NCI NIH Intramural Research Program and NIH Clinical Center Bench-to-Bedside Award.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Paola Gasperinin and Giovanna Tosato for their help with the apoptosis experiments.
- Received March 9, 2007.
- Revision received May 2, 2007.
- Accepted May 10, 2007.
- ©2007 American Association for Cancer Research.