Cellular Metabolism of 2’,3’-Dideoxycytidine, a Compound Active against Human Immunodeficiency Virus in Vitro*

The nucleoside analog 2’,3’-dideoxycytidine (ddcyd) has been shown to inhibit the infectivity and cytopathic effect of human immunodeficiency virus on human OKT4+ lymphocytes in vitro (1). Metabolism of ddCyd by human T-lymphoblastic cells (Molt 4) negative for human immunodeficiency virus and OKT4 was exam- ined. Molt 4 cells accumulated ddCyd and its phosphorylated derivatives into acid-soluble and acid-insoluble material in a dose-dependent manner. For each concentration tested, 2‘,3’-dideoxycytidine triphosphate rep- resented 40% of the total acid-soluble pool of ddCyd metabolites. Uptake of 5 PM ddCyd was linear for 4 h after addition of drug. Efflux of ddCyd metabolites from cells followed a biphasic course with an initial retention half-life of 2.6 h for 2’,3‘-dideoxycytidine triphosphate. DNA, but not RNA, of cells incubated with [‘HIddCyd became radiolabeled. Nuclease and phosphatase treatment of DNA followed by reverse-phase high pressure liquid chromatography showed that the nucleoside was incorporated into DNA in its original form. ddCyd was not susceptible to deamina- tion by human Cyd-dCyd deaminase. It was a poor substrate for human cytoplasmic and mitochondrial dCyd kinases, with K,,, values of 180 f 30 and 120 2


ied in their sensitivity to inhibition by ddCTP with
Ki values of 110 f 40, 2.6 f 0.3, and 0.016 f 0.008 PM, respectively; however, inhibition was competitive with dCTP in each case.
Acquired immunodeficiency syndrome (AIDS)' has emerged as a major health threat in the last several years. Human immunodeficiency virus (HIV) is now recognized as the etiological agent of this disease (2-8). This virus preferentially infects and destroys OKT4' (helper inducer) T-lymphocytes (6). In AIDS and its preceding lymphadenopathy syndrome, there is detectable virus replication as determined by the presence of particulate reverse transcriptase activity (2). This raises the possibility of preventing or slowing the progress of the disease by early detection and blocking viral *This work was supported in part by Grant CH-29 from the American Cancer Society and Grant CA-27448 from the National Institutes of Health. 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.
replication with specific inhibitors of HIV reverse transcriptase.
It was recently reported that 2',3'-dideoxypurine and -pyrimidine nucleosides inhibited the infectivity and cytopathic effect of HIV toward OKT4' cells in vitro at concentrations that did not inhibit proliferation in response to T cell mitogens (1). ddCyd was the most potent of the compounds tested and completely blocked the viral cytopathic effect at concentrations greater than 0.5 PM, a dosage at least 10-fold less than that which inhibited cell growth in the absence of virus.
The activity of ddCyd against HIV and its low cytotoxicity make it an attractive candidate for clinical trial; therefore, it is important to study its metabolism in human cells. We have chosen a T-lymphoblastic cell line (Molt 4) for these studies and report here on the uptake, efflux, phosphorylation, deamination, and incorporation into DNA of ddCyd.

RESULTS
Uptake of ddCyd-Molt 4 cells were incubated with 2,5, or 10 PM [3H]ddCyd for 6 h, and the uptake into acid-soluble and acid-insoluble fractions was analyzed. As seen in Table  1, there was a concentration-dependent uptake into both fractions, although incorporation of [3H]ddCyd into the acidinsoluble pool was only a small percentage of the total. When the acid-soluble fractions were analyzed by anion exchange HPLC (Fig. l ) , there were three major metabolites in addition to a peak which coeluted with a ddCyd marker and an unidentified "solvent breakthrough" peak. The metabolite which eluted in fraction 37 was in a position identical to authentic ddCTP. Metabolites of ddCyd which eluted in fractions 11 and 23 were identified as ddCMP and ddCDP, respectively, on the basis of their elution positions with respect to dCMP and dCDP and by identification of the original nucleoside (ddCyd) after phosphatase treatment (see below). The amount of each metabolite found in cells was dependent on the extracellular concentration of ddCyd, and the proportion of the total acid-soluble radioactivity represented by each metabolite remained the same. In each case, ddCMP, ddCDP, and ddCTP represented approximately 30, 10, and 40%, respectively, of the total acid-soluble radioactive pool. When the acid-soluble fraction from cells treated with 10 PM ddCyd was treated with alkaline phosphatase and venom phosphodiesterase, the radioactivity associated with each metabolite decreased, and there was a corresponding increase in radioactivity associated with [3H]ddCyd. A portion of this phosphatase-treated extract was analyzed on a reverse-phase CIS column, and more than 90% of the radioactivity coeluted with authentic ddCyd (data not shown). This system completely separates ddCyd, which elutes at 10.4 min from Cyd and dCyd, which elute at 4.5 and 5.3 min, respectively.
Uptake and Efflux of ddCyd with Time-Cells were incubated for 2, 4, or 6 h with 5 p~ ddCyd, a concentration approximately equal to the ID50 in this cell line (data not shown). The uptake of ddCyd and formation of its phosphorylated metabolites were linear with time for 4 h after addition of drug ( Fig. 2A). Efflux of ddCyd and its metabolites after resuspension of cells in drug-free medium followed a biphasic course, with an initial retention half-life for the triphosphate of 2.6 h (Fig. 2 B ) . When all phosphory~ated metabolites of ddCyd were included in this calculation, the initial retention half-life was 5.7 h.
I~o r~r a t i o n of ddCyd into Nucleic Acidt+"urified nucleic acids from Molt 4 cells treated for 12 h with 5 p~ 13H]ddCyd were subjected to ultracentrifugation in a Cs2S04 gradient. There was one major peak of radioactivity which corresponded to the position of DNA in the gradient (Fig. 3). It is uncertain at this time whether or not the small amounts of radioactivity associated with RNA represent true incorporation, and more sensitive techniques must be employed to address this question. Nucleic acids digested with micrococcal nuclease, venom phosphodiesterase, and alkaline phosphatase were analyzed on a reverse-phase CIS HPLC column. The majority of the radioactivity eluted in a position identical to that of authentic ddCyd (Fig. 4). The amount and position of radioactive material which eluted before ddCyd varied in different experiments and were most likely due to incomplete digestion of nucleic acids. Susceptibility of ddCyd to Deamination-Since only three major phosphorylated metabolites of ddCyd were seen on the anion exchange HPLC profiles after incubation of cells with ddCyd, it was expected that ddCyd would not be significantly susceptible to deamination. That this was indeed the case is shown in Fig. 5A. When 0.4 mM ddCyd was incubated with 10 units/ml human Cyd-dCyd deaminase for periods up to 90 min, there was no significant catabolism of this compound. In addition, as seen in Fig. 5B, ddCyd interfered only slightly with deamination of dCyd. Deamination of 100 p~ dCyd was 90% of control in the presence of 1 mM ddCyd after 30 min.
Phosphorylation of ddCyd-As shown in Table 2, ddCyd was a poor substrate for both cytoplasmic and mitochondrial dCyd kinase activities, with K, values of 180 2 30 and 120 2 20 pM, respectively. In addition to a high K,,, value, the maximum rate of ddCyd phospho~lation was significantly less than that for dCyd with both enzymes.
Inhibition of Human DNA Polymerases-The three major human DNA polymerases varied widely in their susceptibility to inhibition by ddCTP; however, inhibition was competitive with dCTP for each enzyme. Polymerase y was by far the most sensitive, with a h: value 19-fold less than the Km value for dCTP. Polymerase CY was quite resistant to inhibition, with a Ki value for ddCTP greater than 100 times its K, for dCTP. Polymerase @ was intermediate in sensitivity with a KJKi ratio of 1:l.i'. K; values for ddCTP against DNA polymerases CY, 6, and y were 110 A 40, 2.6 & 0.3, and 0.016 A 0.008 p~, respectively.

DISCUSSION
The data presented show that significant amounts of ddCTP are formed in Molt 4 cells, with the uptake and phosphorylation of ddCyd dependent on its extracellular concentration. ddCTP is a substrate for at least one of the mammalian DNA polymerases since it is incorporated into DNA. Whether or not the small amounts of incorporation can account for the cytotoxicity of ddCyd is under investigation. The relative sensitivities of DNA polymerases a, @, and y to inhibition by ddCTP are similar to those noted by other investigators with ddTTP (14-16) and indicate that polymerases @ and/or y are the most likely targets for inhibition. The observation that polymerase CY, the major enzyme involved in cellular DNA replication (15), is quite resistant to inhibition by ddCTP may explain the relatively low cytotoxicity of ddCyd.
Disappearance of ddCyd metabolites from cells followed a biphasic course, with an initial retention half-life for the triphosphate of 2.6 h. This value is less than the 4-h half-life reported previously for 1-P-D-arabinofuranosylcytosine triphosphate and approximately equal to the half-life of ~-P-Darabinofuranosyi-5-azacytosine triphosphate in Molt 4 cells (10). The biphasic efflux phenomenon may be related to the concentration dependence of enzymatic dephosphorylation of ddCTP, such that when the ddCTP concentration nears the K,,, value for this process, a slower rate of degradation occurs. It should be noted that ddCTP accumuiat~on will be determined not only by the rate of dephosphorylation, but also by the rates of phosphorylation and incorporation into nucleic acids; therefore, it is conceivable that different cell types may vary greatly in their ability to accumulate ddCTP and thus in their susceptibility to ddCyd cytotoxicity.
The studies with human Cyd-dCyd deaminase show that ddCyd is not significantly susceptible to deamination, a major pathway for inactivation of 1-P-D-arabino~ranosylcytosine in humans (17).
The greater potency of ddCyd compared to other dideoxynucleosides to protect against the cytopathic effect of HIV (1) could be due to differences in phosphorylation of these compounds if inhibition of reverse transcriptase by the triphosphate is their mechanism of action as postulated. However, the amounts of phosphorylated ddCyd derivatives formed in cells is small compared to formation of phosphorylated metabolites of other dCyd analogs (10). Indeed, as shown here, ddCyd is a very poor substrate for both kinases responsible for monophosphorylation of dCyd. It is not certain at this time which dCyd kinase, cytoplasmic or mitochondrial, is responsible for the majority of ddCyd monophosphorylation in Molt 4 cells. There is no evidence at this time that HIV induces any unique nucleoside-metabolizing enzymes; therefore, it is likely that one of these kinases is responsible for phosphorylation of ddCyd in HIV-infected cells.
Unpublished results3 from this laboratory using homopolymer primer/template combinations and purified HIV reverse transcriptase indicate that ddTTP and ddGTP are potent inhibitors of this enzyme with Ki values in the nanomolar range. It is unfortunate that inhibition by ddCTP cannot be examined using this system since HIV reverse transcriptase will not use a poly(dG)-oligo(dC) template; however, it is likely that the enzyme has a low Ki value for ddCTP as well.
The small amounts of ddCTP formed in cells may be sufficient to inhibit the viral reverse transcriptase without much effect on cellular DNA polymerases, thus explaining the selective antiviral effect.