MRP4: A previously unidentified factor in resistance to nucleoside-based antiviral drugs.

Dideoxynucleosides, which are potent inhibitors of HIV reverse transcriptase and other viral DNA polymerases, are a common component of highly active anti-retroviral therapy (HAART) (ref. 1). Six reverse transcriptase inhibitors have been approved for human use: azidothymidine; 2'3'-dideoxycytidine; 2'3'-dideoxyinosine; 2', 3'-didehydro-3'deoxythymidine; 2',3'-dideoxy-3'-thiacytidine; and 4-[2-amino-6-(cyclopropylamino)-9H-purin-9-yl]-2-cyclopentene-1-++ +metha nol. Although drug-resistant HIV strains resulting from genetic mutation have emerged in patients treated with HAART (ref. 1), some patients show signs of drug resistance in the absence of drug-resistant viruses. In our study of alternative or additional mechanisms of resistance operating during antiviral therapy, overexpression and amplification of the MRP4 gene correlated with ATP-dependent efflux of PMEA (9-(2-phosphonylmethoxyethyl)adenine) and azidothymidine monophosphate from cells and, thus, with resistance to these drugs. Overexpression of MRP4 mRNA and MRP4 protein severely impaired the antiviral efficacy of PMEA, azidothymidine and other nucleoside analogs. Increased resistance to PMEA and amplification of the MRP4 gene correlated with enhanced drug efflux; transfer of chromosome 13 containing the amplified MRP4 gene conferred resistance to PMEA. MRP4 is the first transporter, to our knowledge, directly linked to the efflux of nucleoside monophosphate analogs from mammalian cells.

Selection of 9-beta-D-arabinofuranosyladenine-resistant human T-lymphoblasts with altered ribonucleotide reductase activity.

We sought to define the cellular activity that mediates resistance in human leukemic cells (CCRF-CEM) to the nucleoside 9-beta-D-arabinofuranosyladenine (ara-A). Stable mutants were obtained by continuous selection at ara-A concentrations of 1 or 2.5 microM in the presence of the adenosine deaminase inhibitor 2'-deoxycoformycin. Four clones selected for further investigation were 4- to 11-fold less sensitive to the cytotoxicity of ara-A than the parental CCRF-CEM line. These clones also showed cross-resistance to deoxyadenosine and thymidine, but normal sensitivity to arabinosylcytosine and adenosine, and increased sensitivity to the etoposide VP16-213. No change was found in the activity of kinases that phosphorylate ara-A and the various nucleosides that could account for the resistant phenotype in these mutant lines. Resistance was associated with a 2- to 8-fold increase in the level of all four deoxyribonucleoside triphosphates. The triphosphate pools in the mutants were resistant to the inhibition produced in wild-type cells by addition of deoxy-adenosine or thymidine, although significant activation in the deoxyguanosine triphosphate pool was obtained by higher concentrations of thymidine. An examination of ribonucleotide reductase in extracts of two of the mutants revealed a specific alteration in the normal sensitivity of the enzyme for deoxyadenosine triphosphate and adenosine triphosphate but not 9-beta-D-arabinofuranosyladenine 5'-triphosphate. When the level of ribonucleotide reductase activity was measured, it was found that the ara-A-resistant cells contained approximately twice the wild-type level of cytidine diphosphate reductase activity at physiological adenosine triphosphate level. This combination of increased enzyme activity and alteration in sensitivity to the nucleoside triphosphates could account for both the changes in deoxyribonucleotide pool sizes and the resistant phenotype of the presumed mutants.

Metabolic basis of arabinonucleoside selectivity for human leukemic T- and B-lymphoblasts.

Purine analogues are potentially useful agents for selective chemotherapy of lymphoproliferative diseases. We compared the toxic effects of various arabinonucleosides against eight human T- and B-lymphoblastoid lines. The arabinosides of cytosine (ara-C), 2-fluoroadenine (F-ara-A), adenine (ara-A) and guanine (ara-G) all inhibited the growth of T-lymphoblasts at concentrations below 2 microM. Only ara-G showed strong selectivity for T-cells, as indicated by a 15- to 250-fold greater toxicity toward T-cell lines than B-cell lines. To investigate the biochemical basis for ara-G selectivity, we compared the metabolism of the arabinonucleosides in CCRF-CEM (T-) versus PF-2S (B-) lymphoblasts. Comparison of arabinonucleoside triphosphate accumulation indicated differences favoring selective ara-GTP formation in T-cells. In contrast, ara-C, ara-A, and F-ara-A formed almost corresponding amounts of their triphosphates in both cell types. Triphosphate accumulation correlated directly with inhibition of DNA synthesis in CCRF-CEM and PF-2S cells. PF-2S cells accumulated less than 20% ara-GTP from the nucleoside than did CCRF-CEM cells. Nucleoside kinase measurements showed no significant differences in arabinonucleoside phosphorylation that could account for the preferential ara-GTP accumulation in T-cells. After removal of arabinonucleoside-containing medium, ara-GTP levels in PF-2S cells declined with a half-life of 49 min whereas, in CCRF-CEM cells, the level of analogue triphosphate remained unchanged. Furthermore, the half-life of ara-CTP, ara-ATP, and F-ara-ATP in the B-cells was 3- to 5-fold longer than that of ara-GTP. These results indicate that ara-G is more selective than other known arabinonucleosides; such selectivity warrants further assessment of the therapeutic potential of this agent against T-cell malignancies and other lymphoid disorders.

Identification of the mechanism of activation of 9-beta-D-arabinofuranosyladenine in human lymphoid cells using mutants deficient in nucleoside kinases.

The biochemical basis of cellular resistance to 9-beta-D-arabinofuranosyladenine (ara-A) and its natural purine derivative, deoxyadenosine, was investigated with two mutants of cultured human T-lymphoblastoid CCRF-CEM cells. One mutant that lacked deoxycytidine kinase activity, designated CEM/ara-C, retained about 10% of wild-type deoxyadenosine kinase and deoxyguanosine kinase activity each but maintained normal adenosine kinase or thymidine kinase activity. This suggested that in these human T-lymphoblastoid cells, as in other previously studied mammalian cells, deoxycytidine and purine deoxyribonucleosides are phosphorylated by the same enzyme. Despite this extensive reduction of purine nucleoside kinase activities, the cytotoxicity of ara-A or deoxyadenosine was not appreciably affected, decreasing by only 2.5- and 6-fold, respectively. A second mutant, isolated by selecting CEM/ara-C mutants that were resistant to ara-A, showed a 100-fold increase in resistance to ara-A cytotoxicity. This ara-A-resistant subline was deficient in the activities of two enzymes, deoxycytidine kinase and adenosine kinase, and showed a high degree of resistance to deoxyadenosine, adenosine, and pyrazofurin but not to pyrimidine analogs, such as 5-fluorodeoxyuridine or 5-fluorouridine. Further studies of ara-A and deoxyadenosine phosphorylation in wild-type and resistant cell lines disclosed that, although deoxycytidine kinase is the principal enzyme for their phosphorylation in vitro, their intracellular conversion to cytotoxic nucleotides depends on the joint action of deoxycytidine kinase and adenosine kinase rather than purine-specific deoxynucleoside kinase, as previously thought.

Tiazofurin metabolism in human lymphoblastoid cells: evidence for phosphorylation by adenosine kinase and 5'-nucleotidase.

The exact route of metabolism of tiazofurin, a novel nucleoside with antitumor activity, is controversial. Using human cell lines severely deficient in salvage nucleotide enzymes, we were able to identify the route of activation in tiazofurin metabolism. With loss of adenosine kinase activity by mutation in two lymphoblastoid cell lines, CCRF-CEM and WI-L2, the growth sensitivity to tiazofurin decreased by 6- and 3-fold, respectively. In contrast, the mutant lines were about 3000- to 1500- and 16- to 4-fold more resistant to the structurally similar tiazofurin analogues pyrazofurin and ribavirin, respectively. Other mutants with defective deoxycytidine or uridine kinase activity showed normal sensitivity to all three analogues. Both cell lines with defective adenosine kinase activity accumulated about 50% wild-type levels of tiazofurin-5'-monophosphate and thiazole-4-carboxamide adenine dinucleotide analogue of tiazofurin at cytotoxic concentrations of the drug. Extracts of wild-type lymphoblasts catalyzed the phosphorylation of tiazofurin in the presence of adenosine 5'-triphosphate and Mg2+. Loss of adenosine kinase activity in the mutant extract eliminated this phosphorylating activity for tiazofurin consistent with the notion that adenosine kinase catalyzes phosphorylation of tiazofurin. However, an enzyme activity that catalyzed the phosphorylation of tiazofurin in the presence of inosine-5'-monophosphate as donor and Mg2+ was detected in the extracts of both wild-type cells and adenosine kinase-deficient mutants. The monophosphate donor specificity, divalent metal, high salt requirement, and nucleoside acceptor specificity of this enzyme activity paralleled that of a 5'-nucleotidase (EC 3.1.3.5) which catalyzes inosine phosphorylation. In addition, tiazofurin phosphorylation was competitively inhibited by inosine and the apparent Ki value was similar to the apparent Km value for inosine phosphorylation. These results indicate that two enzymes, adenosine kinase and a cytoplasmic 5'-nucleotidase, are functionally important anabolizing enzymes for tiazofurin in human cells.

Effects of the epipodophyllotoxin VP-16-213 on cell cycle traverse, DNA synthesis, and DNA strand size in cultures of human leukemic lymphoblasts.

Serial studies of human leukemic lymphoblasts (CCRF-CEM line) cultured with 0.25 to 2.5 microM VP-16-213 for 0 to 6 hr indicated that the mechanism of cytotoxicity of this compound involves a primary effect on DNA. The most striking early change shown by flow cytometry in VP-16-213-treated cells was a delay in S-phase transit before arrest of cells in G2. Coinciding with this S-phase delay was a selective inhibition of thymidine incorporation into DNA as well as concentration-dependent scission of DNA strands. Using alkaline elution methods, we were able to detect DNA breakage at concentrations of VP-16-213 well below the level required to demonstrate kinetic effects or inhibition of DNA synthesis. These data suggest that DNA strand scission is the initial event in the sequence of kinetic and biosynthetic changes leading to growth inhibition and death of VP-16-213-treated cells. Inhibition of replicon initiation due to strand scission is a plausible explanation for the cytotoxic action of this podophyllotoxin derivative.

Growth and differentiation of a human T-cell leukemia cell line, CCRF-CEM, grafted in mice.

The growth of human CCRF-CEM T-cell lymphoblastic leukemia was studied in mice immune deprived by different techniques, and in CD-nu/nu athymic mice. Female CBA/CaJ mice were immune deprived by infant thymectomy, priming with 1-beta-D-arabinofuranosylcytosine (200 mg/kg) 48 h prior to total body irradiation (925 cGy) designated theta ara-C gamma; or after thymectomy the mice received 925 cGy total body irradiation with marrow reconstitution (4 x 10(6) nucleated cells), designated theta gamma BM. Only in mice immune deprived by theta gamma BM, subsequently given a single dose of cyclophosphamide (100 mg/kg) 18-24 h before transplantation of CCRF-CEM, was there progressive reproducible engraftment and tumor growth. For mice immune deprived in this manner the tumor engraftment rate was 100 and 80% of tumors achieved greater than or equal to 1 cm3 within 46 days. In immune-deprived CBA/CaJ mice, but not CD-nu/nu athymic mice, tumor transplanted to the s.c. site metastasized to paraaortic and axillary nodes. Metastatic spread to lymph nodes was confirmed by immunophenotyping and by karyotyping. In contrast to the CCRF-CEM cells in culture, which expressed cytoplasmic CD3 (T3) but not surface CD3, both s.c. and metastatic CCRF-CEM line was exposed to phorbol-12-myristate 13-acetate in vitro to mimic the apparent differentiation which occurred in the xenografted cells, and a similar expression of surface CD3 after treatment was seen. This surface expression of CD3 was accompanied by production of mRNA for the T-cell receptor alpha chain and surface expression of the T-cell receptor. Identical T-cell receptor beta and gamma chain gene rearrangements were found for the CCRF-CEM line in vitro and the xenografted cells in vivo, demonstrating that only one clone was present and that differences in immunophenotyping were not the result of clonal selection. These results suggest that host (mouse) hematopoietic factors could affect human leukemic cell differentiation.

Mode of action of 9-beta-D-arabinosyladenine and 1-beta-D-arabinosylcytosine on DNA synthesis in human lymphoblasts.

The effects of 9-beta-D-arabinosyladenine (AraAde), 1-beta-D-arabinosylcytosine (AraCyt) and 2'-deoxyadenosine on DNA replication in cultured human lymphoblasts (CCRF-CEM line) were studied by pulse-labeling cells with [3H]thymidine and analyzing the nascent DNA by velocity sedimentation in alkaline sucrose gradients. At doses that reduced the overall rate of DNA synthesis to 50--70% of control values, both AraAde and AraCyt profoundly inhibited the formation of new replicons, with secondary effects on chain elongation contributing to the total inhibition of DNA synthesis. In contrast, the suppression of DNA synthesis by 2'-deoxyadenosine stemmed mainly from an inhibition of chain elongation. These studies also disclosed that about 100 times more AraAde than AraCyt was required to produce a similar inhibition of DNA replication in CCRF-CEM cells. Determination of intracellular concentrations of the nucleoside triphosphates (AraCTP and AraATP) indicated to 90% inhibition of DNA synthesis was achieved at 1.6 and 25 pmol/1 . 10(6) cells, respectively. Studies with cell lysates revealed that the replicative machinery in CCRF-CEM cells is more sensitive to AraCTP than to AraATP. This finding contrasts with earlier research, in which the inhibtion of purified DNA polymerase by either AraATP of AraCTP yielded essentially the same Ki value. The difference in sensitivity of the cell lysate to these arabinonucleotides may reflect either a target enzyme other than DNA polymerase or, more plausibly, some subtle interaction of the polymerase with other components of the replicative process.

Cellular factors for resistance against antiretroviral agents.

Substantial advancements have been made in our understanding of the complex replication cycle of, and immunopathology associated with HIV infection as well as the drugs used to treat the disease. The nucleoside reverse transcriptase inhibitors remain the cornerstones of current antiviral treatment modalities. Unfortunately, their longterm use often leads to adverse reactions and the emergence of virus mutants with decreased susceptibility to therapeutic agents. In addition to viral resistance, prolonged antiviral treatment may affect metabolic changes in the host cells that can diminish the efficacy of the treatment. Thus, both viral and cellular resistance mechanisms must be considered in the context of failing antiviral chemotherapy. This review article concerns the intracellular pharmacology of antiviral nucleoside analogues in human lymphoid cells and the possible impact of a newly identified nucleotide transporter on drug resistance.

2'-O-nitro-1-beta-D-arabinofuranosylcytosine. A new derivative of 1-beta-D-arabinofuranosylcytosine that resists enzymatic deamination and has antileukemic activity.

To overcome the susceptibility of the anticancer drug 1-beta-D-arabinofuranosylcytosine (ara-C) to enzymatic deamination, and hence deactivation, we prepared the 2'-O-nitro-1-beta-D-arabinofuranosylcytosine (termed nitrara-C) and evaluated it for biological activity. Nitrara-C was resistant to enzymatic deamination and inhibited the proliferation of several strains of human leukemic T and B lymphoblasts grown in culture. Moreover, it substantially extended the life spans of mice with L1210 leukemia. Studies with ara-C-resistant human leukemic lymphoblasts deficient in deoxycytidine kinase activity disclosed that the inhibitory activity of the new compound depends on its phosphorylation.

Mechanism of uptake of the phosphonate analog (S)-1-(3-hydroxy-2-phosphonylmethoxypropyl)cytosine (HPMPC) in Vero cells.

The cellular uptake of phosphonylmethoxypropyl cytosine (HPMPC) was characterized to gain insight into the molecular properties that allow this anticytomegalovirus drug to permeate cell membranes. The time course of uptake of HPMPC into Vero cells was linear between 10 and 75 min and proportional to the concentration in the medium from 10(-6) to 10(-2) M. HPMPC uptake was temperature sensitive and the rate of uptake was considerably lower at 27 degrees than at 37 degrees and almost totally inhibited at 4 degrees. In competition studies with naturally occurring nucleosides, nucleotides or the phosphonylmethoxyethyl derivatives, none affected the uptake of HPMPC at concentrations up to 2000-fold molar excess. The uptake of [3H]HPMPC into Vero cells was compared with that of [14C]sucrose, a probe for fluid-phase endocytosis. Kinetics for both compounds were very similar, as were the effects of the microtubule antagonist colchicine and the tumor promoting agent phorbol myristate acetate. Colchicine and the phorbol ester are known to, respectively, inhibit and stimulate endocytosis. It is concluded from these data that HPMPC enters Vero cells by fluid-phase endocytosis and that once internalized it may accumulate in the lysosome. Protonation of the negative charge on the phosphonyl group in HPMPC may allow its diffusion across the lysosome membrane and eventual activation to its putative active diphosphorylated form in the cell cytoplasm.

Regulation of purine deoxynucleoside phosphorylation by deoxycytidine kinase from human leukemic blast cells.

The kinetics and regulation of nucleoside phosphorylation by highly purified human deoxycytidine kinase from leukemic lymphoblasts were studied. The phosphorylation of purine nucleosides by this enzyme showed sensitivity to the endogenous inhibitors dCTP and UDP three times greater than the phosphorylation of dCyd. Examination of nucleotide pools in human T and B lymphoblasts disclosed that the levels of dCTP and UDP in these cells were sufficient to regulate kinase activity. The enhanced sensitivity of the kinase to dCTP and UDP was related to its reduced ability to interact with purine nucleosides. Comparison of the phosphorylation kinetics for pyrimidine and purine dideoxynucleosides used in antiviral therapy showed that the purine nucleosides were at least 50-fold less efficient as enzyme substrates. These results suggest that the phosphorylation of pharmacologically active purine nucleosides by deoxycytidine kinase is regulated by cellular nucleotide pools.

(S)-1-(3-hydroxy-2-phosphonylmethoxypropyl)cytosine (HPMPC) inhibits HIV-1 replication in epithelial cells, but not T-lymphocytes.

PMEA [9-(2-phosphonylmethoxyethyl)adenine] inhibited both HSV-1 and HIV-1 replication in MT-2 and HeLa-CD4 cells. (S)-1-[3-hydroxy-2-(phosphonylmethoxy)propyl]cytosine (HPMPC) inhibited both these viruses in the epithelioid HeLa-CD4 cells, but did not inhibit either virus in the T-lymphocytic MT-2 cells. PMEA and HPMPC are metabolized to their diphosphorylated forms within cells, which then inhibit viral polymerases. We therefore compared the metabolism of PMEA and HPMPC in MT-2 and HeLa CD4 cells. PMEApp formation was efficient in both the cell types, whereas HPMPCpp levels were approximately 3-10 fold lower in MT-2 cells, compared to HeLa-CD4 cells. These results indicate that HPMPC can inhibt HIV replication in the appropriate cell types, and show that differences in their metabolism cannot account entirely for the lack of antiviral efficacy of HPMPC in MT-2 cells.

Metabolism in human leukocytes of anti-HIV dideoxypurine nucleosides.

Of the dideoxynucleosides described to date, the purine analogues ddA and ddI have exhibited very favorable therapeutic ratios in vitro. ddI is presently undergoing extensive phase I-II clinical trials. Whereas the action of adenosine deaminase (ADA) and purine nucleoside phosphorylase (PNP) is usually to convert a given analogue of Ado to an inactive or less active form, ddI appears to retain the same biological activity as that of the parent ddA. An explanation for these observations was possible when we found that ddI (1) underwent only a slow cleavage to hypoxanthine through the action of PNP and (2) accumulated the same active antiviral metabolite (i.e., ddATP) as ddA in human lymphoid cells. The use of human lymphoid cells with deficiencies in cellular nucleoside kinases and of inhibitors of pathways of nucleotide metabolism have also revealed new aspects of dideoxypurine metabolism in human lymphoid cells, including the identification of a salvage pathway (phosphotransferase/5'-nucleotide pathway) by which ddA/ddI may be metabolized preferentially to the active nucleotide. The effectiveness of ddA and ddI as orally administered antiviral agents may be limited by their susceptibility to acid hydrolysis and the low efficiency for nucleotide conversion in human lymphoid cells. The presence of a fluorine atom in the arabinose configuration on C-2 confers resistance to solvolysis and renders the analogue less susceptible to enzymatic deamination and resistant to phosphorylytic cleavage by PNP. In addition, human lymphoid cells accumulated several fold higher levels of the putative active triphosphate, 2'-F-dd-ara-ATP, than those of ddA or ddI. This increased accumulation of the analogue triphosphate could be accounted for by a more direct conversion of 2'-F-dd-ara-A by a direct phosphorylation through dCyd kinase than ddA. Thus, a single substitution with fluorine at the 2' "up" position of the sugar moiety of ddA markedly improves several biochemical properties relating to dideoxynucleotide accumulation in human lymphoid cells. Whether there are significant alterations of other biochemical properties, such as the ability of the analogue triphosphate to interact with the target enzyme reverse transcriptase, has not yet been determined. Thus, a definitive resolution of the relative merit of ddA/ddI and its 2'-fluoro-arabinosyl analogue is not yet possible on the basis of the studies described here.

Rhodamine 123 and flow cytometry to monitor the cytotoxic actions of nucleoside analogues in nondividing human lymphocytes.

We used human lymphocytes from the peripheral blood of normal individuals to compare the cytolytic actions of 2-chloro-2'-deoxyadenosine (2-Cl-dAdo) and 9-beta-D-arabinofuranosylguanine (ara-G). Membrane integrity was ascertained by flow cytometric quantification of the cells' uptake of the fluorescent mitochondria-specific probe rhodamine 123. Addition of 10 microM ara-G to lymphocyte cultures led to a progressive decline in the number of cells that could assimilate rhodamine 123, and after 2 days exposure to drug less than 50% of the cells showed normal staining compared to untreated cells. 2-Cl-dAdo had similar cytotoxic effects at a 0.1 microM concentration. Under the same conditions trypan blue revealed only a 10-20% toxic effect by these drugs. Using rhodamine 123 in combination with the vital stain propidium iodide, we found that 90% of the cells with abnormal rhodamine 123 uptake also stained with propidium iodide. The changes in membrane permeability to fluorescent dyes correlated with the loss of lymphocyte ability to respond to the plant mitogen phytohemagglutinin. Our data indicate that rhodamine 123 is a more sensitive probe than trypan blue for ascertaining early loss of the membrane integrity of dying lymphocytes. Combined with flow cytometry, rhodamine 123 uptake represents a simple and rapid method for identifying the important biochemical events associated with the cytocidal and cytostatic effect of drugs against normal and leukemic cells.

Cellular phosphorylation of 2',3'-dideoxyadenosine-5'-monophosphate, a key intermediate in the activation of the antiviral agent DDI, in human peripheral blood mononuclear cells.

2',3'-dideoxyadenosine 5-monophosphate (ddAMP), is a key intermediate in the metabolism of the antiviral agent 2',3'-dideoxyinosine (ddI) to its active triphosphate derivative, 2',3'-dideoxyadenosine-5'-triphosphate (ddATP). The potential role of adenylate kinase in the phosphorylation of ddAMP was studied in human peripheral blood mononuclear cells (PBMC) and a human T cell line, CEMss. Subcellular distribution, sulfhydryl inhibitor, and substrate specificity studies support the hypothesis that the mitochondrial adenylate kinase (AK2) is a major route of cellular activation of these compounds in human lymphocytes.