Arabinosylguanine Is Phosphorylated by Both Cytoplasmic Deoxycytidine Kinase and Mitochondrial Deoxyguanosine Kinase1

Home , Deoxyadenosine kinase

[CANCER RESEARCH 62, 3100–3105, June 1, 2002] Arabinosylguanine Is Phosphorylated by Both Cytoplasmic Deoxycytidine Kinase and Mitochondrial Deoxyguanosine Kinase1

Carlos O. Rodriguez, Jr., Beverly S. Mitchell, Mary Ayres, Staffan Eriksson, and Varsha Gandhi2 Departments of Experimental Therapeutics [C. O. R., M. A., V. G.] and Leukemia [V. G.], The University of Texas M. D. Anderson Cancer Center, and The Graduate School of Biomedical Sciences, Houston, Texas 77030-4095; Departments of Pharmacology and Medicine, University of North Carolina-Chapel Hill, Chapel Hill, North Carolina 27599 [B. S. M.]; and Department of Veterinary Medical Chemistry, Swedish University of Agricultural Sciences, The Biomedical Centre, S-751 23 Uppsala, Sweden [S. E.]

ABSTRACT oside transporters (5). The rate-limiting step in the formation of the active form, ara-GTP, is the phosphorylation of ara-G to its mono- ␤ The prodrug of 9- -D-arabinosylguanine (ara-G), nelarabine, demon- phosphate form (ara-GMP) by nucleoside kinase (6). The ara-GMP is strated efficacy against T-cell acute lymphoblastic leukemia, and its ef- converted to di- and triphosphate forms. In vitro investigations using fectiveness correlated with the accumulation of the triphosphate form (ara-GTP). Although in vitro investigations using purified deoxycytidine T-lymphoid, B-lymphoid, or myeloid human leukemia cell lines dem- kinase (dCK) or deoxyguanosine kinase (dGK) suggested that ara-G is a onstrated that the differential accumulation of ara-GTP in these cells substrate for both enzymes, controversy exists in regard to the role of might be responsible for T-cell-selective cytotoxicity (7–10). Lineage- these enzymes in whole cells. In this work, we used a CEM mutant cell line specific accumulation of ara-GTP was also observed in primary -containing low endogenous levels of dGK and deficient in dCK (dCK؊)to leukemia cells obtained from patients (6, 11). Cellular pharmacoki assess the role of these kinases in ara-G phosphorylation. Using a retro- netic studies during a Phase I trial demonstrated that accumulation of ؊ viral vector system, we infected the dCK mutant cell line to obtain cell ara-GTP was higher in circulating T lymphoblasts than in B lympho- ؉ ؉ lines with overexpression of dCK (dCK ) or dGK (dGK ). Only the blasts or myeloblasts (3). Correlation of the cellular pharmacokinetics ␤ ؉ dCK cell line phosphorylated 1- -D-arabinofuranosylcytosine (used as a and clinical response revealed that patients who achieved a complete substrate for dCK) in a cell-free system; phosphorylation of this com- ؉ or partial response accumulated significantly higher peak ara-GTP pound by dGK was below the limit of detection. Again in in vitro assays, the dCK؊ and dCK؉ cell lines phosphorylated dGuo to similar levels levels (3) than did nonresponding patients. These findings clearly -and 0.93 ؎ 0.19 pmol/mg/min, respectively), whereas dGK؉ demonstrate the importance of the intracellular accumulation of ara 0.15 ؎ 0.91) phosphorylated dGuo more efficiently (150 pmol at 60 min). When ara-G GTP in achieving clinical responses to nelarabine therapy. was used as a substrate in a cell-free system, the maximum accumulation Once phosphorylated, the fraudulent nucleoside triphosphate com- of phosphorylated product was observed in dGK؉ extracts at low ara-G petes with the native deoxynucleotide as a substrate for incorporation ؉ levels (10 ␮M) and in dCK extracts at high concentrations of ara-G (100 into DNA by DNA polymerases (12, 13). These polymerases halt at ␮M). Thus, both dCK and dGK can phosphorylate ara-G, but at low the site of analogue incorporation, which results in inhibition of DNA ara-G concentrations, dGK seems to predominate, whereas at higher synthesis and the initiation of programmed cell death (14). The ara-G concentrations, dCK seems to be the preferred enzyme. In whole- incorporation of the analogue into DNA and subsequent apoptotic cell systems after a 3-h incubation with 10 ␮M ara-G, both dCK؉ and response is essential for the antineoplastic activity of clinically tested dGK؉ cells accumulated ara-GTP; however, the levels were significantly and effective nucleoside analogues such as cladribine (15), cytarabine ,higher in dGK؉ cells. In contrast, at 100 ␮M ara-G (0.0008 ؍ P) in these (16), fludarabine (17, 18), and gemcitabine (18, 19), as well as (0.5529 ؍ intracellular ara-GTP accumulated to similar levels (P ؉ ؉ .(cell types; 25 ؎ 3.7 ␮M in dCK , and 27.8 ؎ 2.7 ␮M in the dGK cells. nelarabine (14 These results from whole-cell experiments are consistent with those from As mentioned above, the accumulation of intracellular ara-GTP is the cell-free system and strongly suggest that ara-G is phosphorylated by important for clinical responses. Clinically relevant analogues such as both kinases, and at low substrate concentrations, dGK is preferred cytarabine (20), cladribine (20), fludarabine (20), gemcitabine (21), enzyme. Evaluation of the expression of each of these kinases in primary and decitabine (22) are activated by dCK. However, in the case of leukemia cells may reveal a biochemical basis for the pharmacological ara-G, controversy exists as to which enzyme is responsible for this differences in the accumulation of ara-GTP. critical first step. Nucleoside kinase assays in cell-free systems or in cell lines with dCK or without dCK have demonstrated that ara-G INTRODUCTION serves as a substrate for cytosolic dCK (2, 10, 20). Purified mitochondrial dGK has also been shown to use ara-G with an affinity similar The success of purine and pyrimidine analogues (1) has generated to dGuo (23–25). Until now, no investigations have been done with interest in other nucleoside analogues such as ara-G,3 which is a whole-cell system to address the roles of each of these kinases in congener of dGuo. A Phase I clinical trial of a water-soluble prodrug ara-G phosphorylation. To determine the role of each of these kinases, of ara-G (GW506U78, nelarabine; Ref. 2) demonstrated its efficacy we used a mutant form of T-cell lymphoblastic cell line CCRF-CEM against relapsed hematologic malignancies (3, 4). that was made resistant to ara-C by chronic exposure to ara-C (26). Metabolically, the nontoxic prodrug nelarabine is demethoxylated This cell line lacks dCK activity and has very low level of endogenous by adenosine deaminase to ara-G (2), which permeates the cell via dGK activity. These cells were infected with retrovirus containing nitrobenzylthioinosine-sensitive and -insensitive equilibrative nucle- cDNA of dCK and dGK, and were used to determine the role of each enzyme in ara-G phosphorylation. Received 4/11/01; accepted 3/26/02. 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 MATERIALS AND METHODS 18 U.S.C. Section 1734 solely to indicate this fact. 1 Supported in part by Grant CA57629 from the National Cancer Institute, Department of Health and Human Services. Chemicals and Reagents. Ara-G was purchased from R. I. Chemicals, Inc. 2 To whom requests for reprints should be addressed, at Department of Experimental (Orange, CA). [8-3H]dGuo, [8-3H]ara-G, and [5-3H]ara-C were purchased Therapeutics, Box 71, The University of Texas M. D. Anderson Cancer Center, Houston, TX from Moravek Biochemicals (Brea, CA). Ara-GTP was custom-synthesized by 77030. Phone: (713) 792-2989; Fax: (713) 794-4316; E-mail: [email protected]. 3 Sierra Bioresearch (Tucson, AZ). The abbreviations used are: ara-G, 9-␤-D-arabinosylguanine; ara-C, arabinosylcy- tosine; ara-GMP; ara-G monophosphate; ara-GTP, ara-G triphosphate; dCK, deoxycyti- Cell Line. The CCRF-CEM T-lymphoblast cell line and its Ara-C8D Ϫ dine kinase; dGK, deoxyguanosine kinase; dGuo, deoxyguanosine. derivative (CCRF-CEM, dCK ) cell line have been described previously (26). 3100

The latter cell line contains mutated dck gene with no detectable protein or Measurement of Intracellular Nucleoside Triphosphates by High- functional dCK activity; the dgk gene is unaffected, and functional dGK pressure Liquid Chromatography. To quantify ara-GTP accumulation in activity are present (26). These cell lines were cultured in DMEM-low glucose whole cells, we incubated exponentially growing cells (5 ϫ 106-1 ϫ 107) with supplemented with 10% heat-inactivated fetal bovine serum (Life Technolo- 100 ␮M ara-G. Nucleotides were extracted by perchloric acid, neutralized with gies, Inc., Grand Island, NY) and maintained in mid-log phase growth at 37°C KOH, and analyzed using high-pressure liquid chromatography (32). Cell Proliferation Assay. Briefly, CEM cells were cultured into 96-well in 5% CO2 in a fully humidified incubator. Under these conditions, the cell doubling time was 24 h. Cells were routinely tested for Mycoplasma by using dishes at a concentration of 20,000 cells/well and incubated with ara-G at a kit (Gen-Probe Inc., San Diego, CA). indicated concentrations. After 24-hours of incubation, cell proliferation was Construction of CEM Cell Lines Overexpressing dCK or dGK. The determined using the MTS Cell Titer Aqueous assay, (Promega, Madison, WI), dCK cDNA cloning and expression have been described before (27, 28). The which measured the conversion of a tetrazolium compound into formazan by dGK cDNA was obtained by PCR from polyadenylated RNA isolated from a mitochondrial dehydrogenase enzyme in live cells. The amount of formazan Raji B lymphoblasts. The full-length cDNA (dGKϩ) was obtained by primers was measured spectrophotometrically and was linear with the cell number. Each data point was the average of three independent determinations. dGK-1FW (5Ј-TAGGGATCCGAATCGTGGGAATGG) and dGK-3RV (5Ј- Calculations and Statistical Analysis. Student’s two-tailed unpaired t test CGGGATCCTTACAGATTCTTT). All of the primers were checked for se- was used to determine the significance of differences in ara-GTP accumulation quence analysis as described previously (27). in whole cells between cell lines. The rates of substrate phosphorylation and The Moloney murine leukemia/sarcoma-based retroviral vector LNPO-dCK ara-GTP accumulation were determined by linear regression analysis that has been described (27, 28). LNPO-dGK contained mitochondrial leader included the 0-h time point. The rates of substrate phosphorylation obtained sequence for mitochondrial localization. LNPO-dCK and LNPO-dGK were during in vitro assays (mean with SE) were compared between cell lines by made by inserting the dCK or dGK cDNA into the BamH1 site of the pLNPO two-tailed Student’s t tests. vector. The dCKϪ cell line was infected with either LNPO-dCK or LNPO- dGK (27). Immunoblot Assays. Cells (1 ϫ 107) were lysed in lysis buffer containing RESULTS 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP40, 1 mM sodium vanadate, Expression of dCK and dGK. To verify protein expression of 1mM phenylmethylsulfonyl fluoride, and 1 mM DTT. One protease inhibitor Ϫ dCK and dGK in dCK CCRF-CEM cells after retroviral infection, mixture tablet was added to each 10 ml lysis buffer (Boehringer Mannheim, we performed Western blot analysis of whole cell lysates (Fig. 1). The Manheim, Germany). Cell lysates (50 ␮g protein) were resolved by 10% SDS-PAGE and then electrophoretically transferred onto Immobilon P mem- parental wild-type CEM cells had high endogenous dCK levels and relatively low endogenous dGK contents. The ara-C-resistant cell line branes (Millipore, Bedford, MA). After being blocked with 4% nonfat dry milk Ϫ in PBS-T (PBS with 0.05% Tween 20) for 1 h, the membranes were probed (dCK ) showed little or no expression of the dCK protein but main- with either rabbit polyclonal anti-dCK peptide antibodies (29) or affinity tained endogenous levels of the dGK protein. Given the absence of a Ϫ purified rabbit polyclonal anti-dGK antibodies (30). The membranes were cell line that lacks dGK, we used this dCK cell line to reconstitute ϩ washed in PBS-T and incubated with horseradish peroxidase-conjugated goat dCK and overexpress dGK. Cells containing transfer of dck (dCK ; antirabbit antibodies for 1 h before visualization with an enhanced chemilu- Fig. 1, Lane 3) produced about half of the dCK protein and the same minescence detection system (NEN Life Science Products, Inc., Boston, MA). amounts of dGK as did the parental cell line. Cells infected with dgk The relative expression of dCK and dGK was quantitated using a densi- (dGKϩ) produced little or no dCK (Fig. 1, Lane 4) and about five tometer and normalized to the value obtained for actin within the same samples times the amount of dGK as did the parental cells or the other two Ϫ on the same blot. The amount of dCK found in dCK CEM cells served as the mutant cell lines. control, and that amount was given a value of 1. The background density from Functional Activity of Kinases in Whole-Cell Extracts. To ver- an exposed and developed portion of the film was determined from pixels ify that the transduced kinases were functional, we tested substrates of within an area identical to that containing the bands of interest. This value was dCK (ara-C) and dGK (dGuo) in an in vitro assay system. Because subtracted from the density value of each sample. dCyd is a substrate for both deoxyadenosine kinase and dGK, we used Enzyme Extraction Protocol. Exponentially growing cells (2 ϫ 107) were lysed by three freeze-thaw cycles (31). For dCK, the extraction buffer con- ara-C as a pure substrate for dCK. Extracts from the cell line infected tained 50 mM Tris-HCl (pH 7.6), 20% glycerol, 0.5% NP40, and 2 mM DTT. One protease inhibitor mixture tablet was added for every 10 ml of extraction buffer (Boehringer Mannheim). For dGK, 0.05% Triton X-100 was substituted for the NP40. The lysates were cleared by centrifugation, and the protein concentration was determined by the Bradford method as per manufacturer’s instructions (Bio-Rad Laboratories, Hercules, CA). Enzyme Activity Assays. The functional activity of dCK or dGK was determined as described previously (31). For dCK, 1 ␮M [3H]ara-C was used as the substrate. Deaminase inhibitor, tetrahydrouridine, was added to the reaction mixture. For dGK, 20 ␮M [3H]dGuo was used as the substrate and to prevent its phosphorolysis, 100 ␮M 8-aminoguanosine was included in the reaction mixture. For the relative ara-G kinase activity, reaction mixtures containing either 10 ␮M or 100 ␮M ara-G were used. To compare and inhibit phosphorylation of dGuo or ara-G by dCK, kinase assays were performed in the absence or presence of high concentrations (150 ␮M) of deoxycytidine (the preferred substrate of dCK) in the assay mixture. Aliquots (3 ␮l) of enzyme extract were mixed and the reactions allowed to progress. Aliquots (20 ␮l) of the reaction were spotted on DE-81 filter papers (Whatman, Maidstone, United Kingdom) and allowed to dry. The filter papers Fig. 1. Expression of dCK and dGK proteins in ara-C-resistant CEM cells. Ara-C- were washed in 5-mM ammonium formate, then in ethanol, were air-dried, and resistant CEM (dckϪ) were infected with retrovirus containing dck or dgk. Exponentially the radioactivity was quantitated. Under these conditions, the lower limit of growing cultures of parental CCRF-CEM (Lane 1), dCKϪ (Lane 2), dCKϩ (Lane 3), and ϩ detection was ϳ0.05 nmol of ara-C/mg of protein or 0.05 pmol of dGuo or dGK (Lane 4) were harvested, and whole-cell extracts were prepared, resolved by SDS-PAGE, and transferred to nylon membranes, after which the membranes were probed ara-G/mg of protein. The results were expressed as nmol (for dCK) or pmol for dCK (top blot), dGK (middle blot), or actin (bottom blot). Experiments were repeated (for dGK) of phosphorylated product per mg protein. three times with similar results. 3101

Fig. 2. In vitro phosphorylation of ara-C or dGuo. Extracts of dCKϪ cells (Ⅺ), dCKϩ (f), and dGKϩ cells (F) were prepared and used as described in “Materials and Methods.” The phosphorylation of ara-C (A) or dGuo in the absence (B)or presence (C) of deoxycytidine was quantified as a function of time and expressed as nmol/mg protein or pmol/mg protein, respectively. Data points are the means of three separate experiments conducted in quadruplicate; bars, Ϯ SE. Where the bars are not visible, they are contained within the symbol.

with dck (dCKϩ) phosphorylated [3H]ara-C at a constant rate P ϭ 0.34; n ϭ 4). An identical pattern of ara-G phosphorylation was 2 (1.0 Ϯ 0.03 nmol/mg protein/min; r ϭ 0.99; Fig. 2A). This increase observed when 1 ␮M ara-G was tested (data not shown). in phosphorylation was because of exogenously expressed dCK rather To test the relative importance of dCK and dGK at clinically than endogenous dGK, because dCKϪ cells could not phosphorylate relevant ara-G concentrations, we repeated the experiments using a 3 ␮ ara-C to detectable levels under similar conditions. As expected, [ H]ara-G concentration of 100 M, which is near the Km of dCK for ϩ overexpression of dGK in the absence of dCK did not result in any ara-G but 2–10 times the Km of dGK for ara-G. At this substrate phosphorylated products of ara-C (Fig. 2A). concentration, rates of ara-G phosphorylation were similar for dCKϪ Similarly, to determine the in vitro kinase activity of dGK in these (4.13 Ϯ 0.43 pmol/mg/min) and dGKϩ cells (4.59 Ϯ 0.34 pmol/mg/ cell lines, we also tested whole-cell extracts for their ability to min; Fig. 3C). Again, the addition of dCyd (Fig. 3D) had no effect on phosphorylate the substrate [3H]dGuo. We confirmed that all three of the rate of ara-G phosphorylation in either dCKϪ (4.12 Ϯ 0.20 3 ϩ the cell lines were able to phosphorylate [ H]dGuo in the absence pmol/mg/min) or dGK (4.95 Ϯ 0.37 pmol/mg/min). At 100 ␮M (Fig. 2B) or presence (Fig. 2C) of the dCK substrate, deoxycytidine. ara-G, dCKϩ phosphorylated ara-G at the maximum rate in the The extract prepared from dCKϪ cells, which express endogenous absence of deoxycytidine (24.02 Ϯ 0.67 pmol/mg/min). The addition levels of dGK, phosphorylated [3H]dGuo at a constant rate of of dCyd inhibited the enzyme activity and decreased the rate of ara-G 0.91 Ϯ 0.15 pmol/mg protein/min (r2 ϭ 0.96), a rate similar to that of phosphorylation by ϳ3-fold (9.68 Ϯ 1.60 pmol/mg/min). the reconstituted dCK (dCKϩ) cell line (0.93 Ϯ 0.19 pmol/mg protein/ Whole Cell Accumulation of Ara-GTP. Next, to determine min; r2 ϭ 0.89). Thus for dGuo, dGK is the major phosphorylating whether the expressed enzymes would phosphorylate ara-G in a enzyme at this concentration of substrate. Consistent with that postu- whole-cell milieu, we incubated dCKϪ, dCKϩ, and dGKϩ cells with ϩ late, extracts from the cell line overexpressing dGK (dGK ) produced 100 ␮M ara-G for 3–12 h (Fig. 4). The lowest accumulation of much higher levels of phosphorylated dGuo (Fig. 2, B and C). By 60 min, Ͼ150 pmol of phosphorylated products were formed. The presence of dCyd did not inhibit the dGuo-phosphorylating activity of dGK, additionally confirming that dGuo was the preferred substrate for the dGK enzyme. The results presented above verified that the transduced enzymes were functional with pure substrates. Next, to test the postulate that

dGK phosphorylates ara-G at a Km similar to that of its natural substrate, dGuo, and dCK phosphorylates ara-G with lower affinity, we measured the in vitro ara-G-kinase activity in the transfectants that overexpressed dCK or dGK (Fig. 3). To mimic the kinetic concen- 3 trations, [ H]ara-G was used at a concentration of 10 ␮M (Fig. 3, A ␮ and B), which is near the Km for dGK or at 100 M (Fig. 3, C and D),

which is near the Km for dCK. To compare and determine the role of dCK, the ara-G phosphorylation assay was performed in the absence (Fig. A and C) or presence (Fig. 3, B and D) of the inhibitor deoxycytidine. In experiments with dCK- cells, endogenous levels of dGK could phosphorylate ara-G at the relatively low concentration of 10 ␮M ara-G (Fig. 3A; 0.08 Ϯ 0.01 pmol/mg protein/min; n ϭ 10); the addition of dCyd did not affect this rate (Fig. 3B; 0.10 Ϯ 0.01 pmol/mg/min; n ϭ 8). Extracts from dGKϩ cells phosphorylated ara-G about six times faster (Fig. 3A; 0.48 Ϯ 0.03 pmol/mg protein/ min) than the extracts from dCKϪ cells (P ϭ 0.0237), and again the addition of dCyd did not affect this rate (Fig. 3B; 0.39 Ϯ 0.06 pmol/mg protein/min; n ϭ 8). Moreover, when dCK was restored, the Fig. 3. In vitro phosphorylation of ara-G. Extracts of dCKϪ cells (Ⅺ), dCKϩ cells (f), rate of ara-G phosphorylation was significantly faster (Fig. 3A; ϩ and dGK cells (F) were prepared and analyzed as described in “Materials and Methods.” 0.15 Ϯ 0.01 pmol/mg/min; n ϭ 8) than that of cells containing only The ara-G-phosphorylation activity at 10 ␮M (A and B)or100␮M (C and D) ara-G in the endogenous dGK (P ϭϽ0.0001). As expected, the addition of dCyd absence (A and C) or presence (B and D) of deoxycytidine was quantified as a function ϩ of time and expressed as pmol/mg protein. Data points are the means of at least two inhibited the activity of the transferred dCK , resulting in ara-G being Ϯ Ϫ separate experiments conducted in duplicate; bars, SE. Where the bars are not visible, phosphorylated at the same rate as in dCK cells (0.07 Ϯ 0.02; they are contained within the symbol. 3102

phosphorylation step eventually results in the intracellular accumulation of ara-GTP. When these cell lines were incubated with 100 ␮M ara-G for 3 days, both dCKϩ and dGKϩ cells showed similar (25–30%) inhibition of cell growth suggesting that exogenously expressed kinases were bio- logically functional. This is consistent with previous reports (27, 28).

DISCUSSION As is the case for other arabinosyl analogues such as ara-C (16, 34), the cytotoxicity of ara-G results from the cellular accumulation of Fig. 4. Time-dependent accumulation of ara-GTP in whole cells. Exponentially grow- ara-GTP, the incorporation of which into DNA results in an apoptotic ing cultures of dCKϪ cells (Ⅺ), dCKϩ cells (f), and dGKϩ cells (F) were incubated death signal (14). The initial and rate-limiting step in the accumulation continuously with 100 ␮M ara-G for the indicated times and harvested, after which of ara-GTP from ara-G is the formation of ara-GMP. This phospho- intracellular ara-GTP levels were quantified as described in “Materials and Methods.” Data points are the means of three separate experiments conducted in duplicate; rylation is catalyzed in vitro by high-affinity, low specific activity bars, Ϯ SE. Where the bars are not visible, they are contained within the symbol. mitochondrial dGK (25) and low-affinity, high specific activity cytosolic dCK (2, 20). dGK activity is found in most tissues, including liver, lymphoid tissues such as B and T cells, spleen, skin, and brain Ϫ ara-GTP (Ͻ5 ␮M) was found in dCK cultures, which have low (23, 35–39). Although the mitochondria-rich brain tissue contains endogenous dGK levels. Although there was a time-dependent in- higher specific activity of dGK than other tissues (23), the specific crease in ara-GTP levels, the accumulation reached a plateau in these activity of dGK in all of the tissues is lower than that of dCK. The Ϯ ␮ ␮ cells by 3–6 h. At 12 h, the concentration was 8.0 0.8 M. dGK Km for ara-G is similar to that for its natural substrate (7 M for However, when dGK was overexpressed, the intracellular accumula- dGuo, 7—65 ␮M for ara-G; Refs. 2, 23, 25) suggesting that dGK tion of ara-GTP increased to 48.3 ␮M Ϯ 7.4 ␮M by 12 h (Fig. 4). activity may phosphorylate ara-G at low concentrations but that the Similarly, when dCK was overexpressed the accumulation of ara-GTP phosphorylation would become saturated at higher levels of the sub- also increased and reached 57.7 ␮M Ϯ 16.2 ␮M by 12 h. Moreover, in strate. Our findings in both cell-free and whole-cell systems are both of these cell lines, ara-GTP accumulation did not reach a plateau consistent with this postulate. At low concentrations of ara-G (1 and in the dCK or dGK cells suggesting that longer incubation periods 10 ␮M), the rate of phosphorylation of ara-G and the intracellular would result in continuing accumulation. This observation also indi- accumulation of ara-GTP was greater in dGKϩ cells than in dCKϩ cates that at these concentrations of ara-GTP, the enzymes are not cells (P ϭ 0.01). At higher concentrations of ara-G (100 ␮M), the feedback inhibited. Given the correlation between ara-GTP and re- accumulation of ara-GTP reached a plateau and was similar in dGKϩ sponse to treatment with nelarabine (3), we have explored this phar- and dCKϩ cells. Given the high specific activity of dGK in nerve macological strategy to increase ara-GTP in freshly isolated leukemia cells, it could be speculated that the nervous system might be more cells (6) and designed a protocol to infuse higher levels of nelarabine sensitive to the toxic effects of ara-G. Indeed, grade 2 somnolence and (33). self-limiting peripheral neuropathy were the most common nonhema- Finally, to determine the relationship between ara-G concentrations tologic side effects noted in Phase I trials of nelarabine alone (3) or and ara-GTP accumulation, the three cell lines were exposed to 10 nelarabine combined with fludarabine (40). Similar cytotoxicity might Ϫ ␮M,30␮M, and 100 ␮M ara-G for 3 h (Table 1). In the dCK cell line be expected for leukemic or other neoplastic cells that have naturally the accumulation of ara-GTP at any ara-G concentration was low higher dGK. Ϫ (2.7–6.7 ␮M). Compared with dCK cell line, expression of either A corollary to high expression of dGK is the fact that leukemia cells dGK or dCK resulted in significant accumulation of ara-GTP at all show differential cytotoxicity to nucleoside analogs that are phospho- four of the concentrations of ara-G (see Ps in Table 1 for dCKϪ versus rylated by only dCK compared with that with ara-G. A recent trial of dCKϩ and dCKϪ versus dGKϩ). When dCK was expressed, incuba- fludarabine with nelarabine demonstrated a bimodal accumulation of tion with 10–100 ␮M ara-G resulted in proportional increases in ara-GTP but not fludarabine triphosphate (40) in circulating leukemia ϩ ara-GTP, which ranged from 8.5 Ϯ 1.4 to 35.7 Ϯ 4.5 ␮M.IndGK , cells. The 70-fold variation in the level of ara-GTP found in cells from cells that had no endogenous dCK and overexpressed dGK, 3-h the patients in this trial probably reflected heterogeneity in ara-G incubations with 10 and 30 ␮M ara-G resulted in accumulations of phosphorylation. The variation in accumulation of fludarabine 18.2 Ϯ 1.5 ␮M and 21.7 Ϯ 1.8 ␮M ara-GTP that were higher than triphosphate was much lower (6–7-fold). Because both of these ana- ϩ dCK cells (P ϭ 0.0008 at 10 ␮M; and P ϭ 0.0304 at 30 ␮M). At logues are substrates for dCK, these results cannot be explained by an higher ara-G concentrations, dGKϩ cells accumulated ara-GTP at increase in the specific activity of dCK. It is tempting to speculate, ϩ similar levels as did the dCK cells (P ϭ 0.5529 at 100 ␮M and then, that the population of cells that accumulated more ara-GTP may 0.5659 at 300 ␮M; Table 1). These findings strongly suggest that both be those that expressed a high specific activity of dGK. Prospective dCK and dGK can phosphorylate ara-G in intact cells and that this evaluation of the expression of each of these kinases in primary

Table 1 Concentration-dependent accumulation of ara-GTP in whole cellsa

ara-GTP ␮M, mean Ϯ SE Ps

ara-G, ␮M dCKϪ dCKϩ dGKϩ dCKϪ vs. dCKϩ dCKϪ vs. dGKϩ dCKϩ vs. dGKϩ 10 2.7 Ϯ 0.5 8.5 Ϯ 1.4 18.2 Ϯ 1.5 0.0027 Ͻ0.0001 0.0008 30 2.7 Ϯ 0.6 14.5 Ϯ 2.2 21.7 Ϯ 1.8 0.0005 Ͻ0.0001 0.0304 100 4.0 Ϯ 1.0 25.0 Ϯ 3.7 27.8 Ϯ 2.7 0.0003 Ͻ0.0001 0.5529 300 6.7 Ϯ 1.6 35.7 Ϯ 4.5 39.5 Ϯ 4.7 0.0001 Ͻ0.0001 0.5659 a Data shown are means Ϯ SE of three separate experiments conducted in duplicate. 3103

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2002 American Association for Cancer Research. BOTH dCK AND dGK PHOSPHORYLATE Ara-G leukemia cells may reveal a biochemical basis for the pharmacologic rylated by both mitochondrial dGK and cytosolic dCK. In the absence differences in the accumulation of ara-GTP. These differences may of dCK, the accumulation of ara-GTP was low, maybe because of the identify patients who can benefit from nelarabine as opposed to low endogenous level of dGK. However, when the specific activity of traditional analogs that use only dCK. dGK was increased, the accumulation of ara-GTP was augmented. As mentioned before, the location of endogenous dGK enzyme is in Because dGK is present in mitochondria (30), cells that express the mitochondria (23, 25, 30, 37, 41). In our system, consistent with greater numbers of mitochondria are vulnerable to high accumulation previous report (42), overexpression of dGK results in accumulation of ara-GTP and ara-G-induced cytotoxicity. Similarly, when the spe- of protein in mitochondrial fraction (data not shown). On the basis of cific activity of dCK was amplified, a proportional increase in the cytotoxicity observed with analogs that are phosphorylated by over- accumulation of intracellular ara-GTP was observed. Despite the low expressed dGK, it is postulated that the cells incorporate these analogs affinity of dCK for ara-G, when this enzyme is present, the accumu- and result in cytotoxicity. At present, it is not clear how the triphos- lation of intracellular ara-GTP was also augmented. Finally, these phate of nucleoside analogs, such as ara-G, generated in mitochondria studies illustrate the role of each kinase for phosphorylation of ara-G transport to nucleus for incorporation into DNA to elicit cell death. in intact cells; at low concentrations of the substrate dGK is the Nonetheless, the accumulation of ara-GTP in the cells lacking dCK preferred enzyme; however, at high concentrations of ara-G dCK and overexpressing dGK (dGKϩ), and subsequent cytotoxicity pro- phosphorylates it efficiently. The relative abundance of each kinase in vide an evidence for such a postulate. primary leukemia cells may predict for phosphorylation of ara-G. As is true for dGK, dCK is constitutively expressed in many tissues. However, unlike dGK, the highest specific activity of dCK is found in ACKNOWLEDGMENTS lymphoid tissues, particularly immature T cells (38, 43). In our study, We thank Christine F. Wogan for critically editing the manuscript and a 3-h incubation of parental CEM cells with 100 ␮M ara-G resulted in Debbie Cox for preparation of the manuscript. the intracellular accumulation of ϳ100 ␮M ara-GTP (14), whereas in dCKϪ cells, the loss of this low-affinity high-specific activity enzyme REFERENCES resulted in the intracellular accumulation of only ϳ5 ␮M ara-GTP. Although in dCKϪ cells, the rate-limiting phosphorylation of ara-G 1. Plunkett, W., and Saunders, P. P. Metabolism and action of purine nucleoside analogs. Pharmacol. Ther., 49: 239–268, 1991. reflects the contribution of only dGK, the low accumulated concen- 2. Lambe, C. U., Averett, D. R., Paff, M. T., Reardon, J. E., Wilson, J. G., and tration of ara-GTP underscores the importance of dCK in the dose- Krenitsky, T. A. 2-Amino-6-methoxypurine arabinoside: an agent for T-cell malig- dependent intracellular accumulation of ara-GTP at clinically achiev- nancies. Cancer Res., 55: 3352–3356, 1995. ϩ 3. Gandhi, V., Plunkett, W., Rodriguez, C. O. R., Nowak, B. J., Du, M., Ayres., M., able levels of nelarabine. Because dCK cells contained less dCK Kisor, D. F., Mitchell, B. S., Kurtzberg, J., and Keating, M. J. Compound GW506U78 than did the parental CEM cells, the intracellular accumulation of in refractory hematologic malignancies: relationship between cellular pharmacoki- ϩ netics and clinical response. J. Clin. Oncol., 16: 3607–3615, 1998. ara-GTP in the dCK cells was about five times less than in the 4. Kurtzberg, J., Keating, M. J., Moore, J. O., Gandhi, V., Blaney, S., Gold, S., Ernst, parental cells with unchanged dCK, a finding that additionally T., Henslee-Downey, J., Chang, A., Kisor, D., Plunkett, W., and Mitchell, B. S. strengthens the association between dCK and ara-GTP accumulation. 2-amino-9-␤-D-arabinosyl-6-methoxy-9H-guanine (GW 506: Compound 506U) is Ͼ ␮ highly active in patients with T-cell malignancies: results of a phase I trial in pediatric Because the Km of dCK for ara-G ( 100 M) is much greater than for and adult patients with refractory hematological malignancies. Blood, 88: 669a, 1996. its natural substrate deoxycytidine (Ͻ1 ␮M; Refs. 2, 20), we would 5. Prus, K. L., Averett, D. R., and Zimmerman, T. P. Transport and metabolism of ␤ expect that the dose-dependent phosphorylation of ara-G would be- 9- -D-arabinofuranosylguanine in a human T-lymphoblastoid cell line: nitrobenzylthioinosine-sensitive and -insensitive influx. Cancer Res., 50: 1817–1821, 1990. come saturated only at higher levels of ara-G than those needed to 6. Rodriguez, C. O. R., Legha, J. K., Estey, E., Keating, M. J., and Gandhi, V. saturate dGK. Our findings are consistent with this notion. Further- Pharmacological and biochemical strategies to increase the accumulation of arabinofuranosylguanine triphosphate in primary human leukemia cells. Clin. Cancer Res., 3: more, cells with the highest specific activity of dCK, such as imma- 2107–2113, 1997. ture T cells, would also be expected to be sensitive to the cytotoxic 7. Verhoef, V., and Fridland, A. Metabolic basis of arabinonucleoside selectivity for effects of nelarabine. Indeed, this particular cell type has been shown human leukemic T and B lymphoblasts. Cancer Res., 45: 3646–3650, 1985. 8. Ullman, B., and Martin, D. W. J. Specific cytotoxicity of arabinosylguanine toward to be sensitive to nelarabine in vitro (7–10) and in a Phase I trial of cultured T lymphoblasts. J. Clin. Investig., 74: 951–955, 1984. this prodrug (3). Another cell lineage that has been identified as 9. Cohen, A., Lee, J. W., and Gelfand, E. W. Selective toxicity of deoxyguanosine and possessing high dCK activity is that of B-cell chronic lymphocytic arabinosyl guanine for T-leukemic cells. Blood, 61: 660–666, 1983. 10. Shewach, D. S., Daddona, P. E., Ashcraft, E., and Mitchell, B. S. Metabolism and leukemia. Among patients with a variety of indolent leukemias, pa- selective cytotoxicity of 9-␤-D-arabinofuranosylguanine in human lymphoblasts. Can- tients with B-cell chronic lymphocytic leukemia achieved objective cer Res., 45: 1008–1014, 1985. 11. Shewach, D. S., and Mitchell, B. S. Differential metabolism of 9-␤-D-arabinofurano- responses. The fact that circulating leukemia lymphocytes from these sylguanine in human leukemic cells. Cancer Res., 49: 6498–6502, 1989. patients accumulated high levels of ara-GTP additionally strengthens 12. Ono, K., Ohashi, A., Yamamoto, A., Matsukage, A., Takahasi, T., Saneyoshi, M., and our hypothesis that cells that express high dCK would be particularly Ueda, T. Inhibitory effects of 9-␤-D-arabinofuranosylguanine 5Ј-triphosphate and 9-␤-D-arabinofuranosyladenine 5Ј-triphosphate on DNA polymerases from murine susceptible to nelarabine therapy. Taken together, these data strongly cells and oncornavirus. Cancer Res., 39: 4673–4680, 1979. suggest that in these cells (CEM and patient cells), because of the high 13. Gandhi, V., Mineishi, S., Huang, P., Chapman, A. J., Yang, Y., Chen, F., Nowak, B., specific activity of dCK, dGK may play a minimal role for ara-GTP Chubb, S., Hertel, L. W., and Plunkett, W. Cytotoxicity, metabolism, and mechanisms of action of 2Ј,2Ј-difluorodeoxyguanosine in Chinese hamster ovary cells. Cancer accumulation. Res., 55: 1517–1524, 1995. In contrast to mitochondrial localization of dGK, endogenous na- 14. Rodriguez, C. O. R., and Gandhi, V. Arabinosylguanine-induced apoptosis of T- lymphoblastic cells: incorporation into DNA is a necessary step. Cancer Res., 59: tive dCK is a cytosolic enzyme (29, 44); however, overexpression of 4937–4943, 1999. the protein may result in nuclear localization (29, 42). Irrespective of 15. Hentosh, P., Koob, R., and Blakley, R. L. Incorporation of 2-halogeno-2Ј-deoxyade- the location of overexpressed enzyme, the phosphorylated product nosine 5-triphosphates into DNA during replication by human polymerases ␣ and ␤. J. Biol. Chem., 265: 4033–4040, 1990. accumulates in the cell and incorporates into DNA to initiate cyto- 16. Kufe, D. W., Major, P. P., Egan, E. M., and Beardsley, G. P. Correlation of toxicity. This has been demonstrated both in cell lines (27, 28) and in cytotoxicity with incorporation of ara-C into DNA. J. Biol. Chem., 255: 8997–900, vivo in tumor model system (28). In the present study also, expression 1980. 17. Huang, P., Chubb, S., and Plunkett, W. Termination of DNA synthesis by 9-␤-D- of dCK in the dCK-deficient cell line results in accumulation of arabinofuranosyl-2-fluoroadenine. A mechanism for cytotoxicity. J. Biol. Chem., ara-GTP in the cytosol (Table 1; Fig. 4), and this was associated with 265: 16617–16625, 1990. 18. Huang, P., and Plunkett, W. Fludarabine- and gemcitabine-induced apoptosis: incor- cell death (data not shown). poration of analogs into DNA is a critical event. Cancer Chemother. Pharmacol., 36: In conclusion, we have demonstrated here that ara-G was phospho- 181–188, 1995. 3104

19. Huang, P., Chubb, S., Hertel, L. W., Grindey, G. B., and Plunkett, W. Action of 2Ј, the determination and quantitation of arabinosylguanine triphosphate and fludarabine 2Ј-difluorodeoxycytidine on DNA synthesis. Cancer Res., 51: 6110–6117, 1991. triphosphate in human cells. J. Chromatogr. B Biomed. Appl., 745: 421–430, 2000. 20. Krenitsky, T. A., Tuttle, J. V., Koszalka, G. W., Chen, I. S., Beacham, L. M., Rideout, 33. O’Brien, S., Thomas, D., Kantarjian, H. M., Freireich, E., Koller, C., Giles, F., Bivins, J. L., and Elion, G. B. Deoxycytidine kinase from calf thymus. Substrate and inhibitor C., Lerner, S., Hodge, J. P., and Keating, M. Compound 506 has activity in mature specificity. J. Biol. Chem., 251: 4055–4061, 1976. lymphoid leukemia. Blood, 92: 490a. 1998. 21. Gandhi, V., and Plunkett, W. Modulatory activity of 2Ј,2Ј-difluorodeoxycytidine on 34. Major, P. P., Egan, E. M., Beardsley, G. P., Minden, M. D., and Kufe, D. W. Lethality the phosphorylation and cytotoxicity of arabinosyl nucleosides. Cancer Res., 50: of human myeloblasts correlates with the incorporation of arabinofuranosylcytosine 3675–3680, 1990. into DNA. Proc. Natl. Acad. Sci. USA, 78: 3235–3239, 1981. 22. Momparler, R. L., and Derse, D. Kinetics of phosphorylation of 5-aza-2Ј-deoxycy- 35. Yamada, Y., Goto, H., and Ogasawara, N. Regulation of human placental deoxy- tidine by deoxycytidine kinase. Biochem. Pharmacol., 28: 1443–1444, 1979. guanosine kinase by nucleotides. FEBS Lett., 157: 51–53, 1983. 23. Wang, L., Karlsson, A., Arner, E. S., and Eriksson, S. Substrate specificity of 36. Sarup, J. C., and Fridland, A. Identification of purine deoxyribonucleoside kinases mitochondrial 2Ј-deoxyguanosine kinase. Efficient phosphorylation of 2-chlorode- from human leukemia cells: substrate activation by purine and pyrimidine deoxy- oxyadenosine. J. Biol. Chem., 268: 22847–22852, 1993. ribonucleosides. Biochemistry, 26: 590–597, 1987. 24. Green, F. J., and Lewis, R. A. Partial purification and characterization of deoxy- 37. Park, I., and Ives, D. H. Properties of a highly purified mitochondrial deoxyguanosine guanosine kinase from pig skin. Biochem. J., 183: 547–553, 1979. kinase. Arch. Biochem. Biophys., 266: 51–60, 1988. 25. Lewis, R. A., and Link, L. Phosphorylation of arabinosyl guanine by a mitochondrial 38. Eriksson, S., Arner, E., Spasokoukotskaja, T., Wang, L., Karlsson, A., Brosjo, O., enzyme of bovine liver. Biochem. Pharmacol., 38: 2001–2006, 1989. Gunven, P., Julusson, G., and Liliemark, J. Properties and levels of deoxynucleoside 26. Owens, J. K., Shewach, D. S., Ullman, B., and Mitchell, B. S. Resistance to kinases in normal and tumor cells; implications for chemotherapy. Adv. Enzyme 1-␤-D-arabinofuranosylcytosine in human T-lymphoblasts mediated by mutations Regul., 34: 13–25, 1994. within the deoxycytidine kinase gene. Cancer Res., 52: 2389–2393, 1992. 39. Herrstro¨m Sjo¨berg, A., Wang, L., and Eriksson, S. Antiviral guanosine analogs as 27. Hapke, D. M., Stegmann, A. P., and Mitchell, B. S. Retroviral transfer of deoxycy- substrates for deoxyguanosine kinase: implications for chemotherapy. Antimicrob. tidine kinase into tumor cell lines enhances nucleoside toxicity. Cancer Res., 56: Agents Chemother., 45: 739–742, 2001. 2343–2347, 1996. 40. Gandhi, V., Plunkett, W., Weller, S., Du, M., Ayres., M., Rodriguez, C. O. R., 28. Manome, Y., Wen, P. Y., Dong, Y., Tanaka, T., Mitchell, B. S., Kufe, D. W., and Ramakrishna, P., Rosner, G. L., Hodge, J. P., O’Brien, S., and Keating, M. J. Fine, H. A. Viral vector transduction of the human deoxycycytidine kinase cDNA Evaluation of the combination of nelarabine and fludarabine in leukemias: clinical sensitizes glioma cells to the cytotoxic effects of cytosine arabinoside in vitro and in response, pharmacokinetics and pharmacodynamics in leukemia cells. J. Clin. Oncol., vivo. Nat. Med., 2: 567–573, 1996. 19: 2142–2152, 2001. 29. Hatzis, P., Said Al-Madhoon, A., Ju¨llig, M. Petrakis, T. G., Eriksson, S., and 41. Zhu, C., Johansson, M., Permert, J., and Karlsson, A. phosphorylation of anticancer Talianidis, I. The intracellular localization of deoxycytidine kinase. J. Biol. Chem., nucleoside analogs by human mitochondrial deoxyguanosine kinase. Biochem. Phar- 273: 30239–30243, 1998. macol., 56: 1035–1040, 1998. 30. Jullig, M., and Eriksson, S. Mitochondrial and submitochondrial localization of 42. Johansson, M., Brismar, S., and Karlsson, A. Human deoxycytidine kinase is located human deoxyguanosine kinase. Eur. J. Biochem., 267: 5466–5472, 2000. in the cell nucleus. Proc. Nat. Acad. Sci. USA, 94: 11941–11945, 1997. 31. Spasokoukotskaja, T., Arner, E. S., Brosjo, O., Gunven, P., Juliusson, G., Liliemark, 43. Carson, D. A., Kaye, J., and Seegmiller, J. E. Differential sensitivity of human J., and Eriksson, S. Expression of deoxycytidine kinase and phosphorylation of leukemic T cell lines and B cell lines to growth inhibition by deoxyadenosine. 2-chlorodeoxyadenosine in human normal and tumour cells and tissues. Eur. J. J. Immunol., 121: 1726–1731, 1978. Cancer, 31A: 202–208, 1995. 44. Gower, W. R., Jr., Carr, M. C., and Ives, D. H. Deoxyguanosine kinase. Distinct 32. Rodriguez, C. O. R., Plunkett, W., Paff, M. T., Du, M., Nowak, B., Ramakrishna, P., molecular forms in mitochondria and cytosol. J. Biol. Chem., 254: 2180–2183, Keating, M. J., and Gandhi, V. High-performance liquid chromatography method for 1979.

3105

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2002 American Association for Cancer Research. Arabinosylguanine Is Phosphorylated by Both Cytoplasmic Deoxycytidine Kinase and Mitochondrial Deoxyguanosine Kinase

Carlos O. Rodriguez, Jr., Beverly S. Mitchell, Mary Ayres, et al.

Cancer Res 2002;62:3100-3105.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/62/11/3100

Cited articles This article cites 44 articles, 28 of which you can access for free at: http://cancerres.aacrjournals.org/content/62/11/3100.full#ref-list-1

Citing articles This article has been cited by 12 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/62/11/3100.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/62/11/3100. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.