L- and D-enantiomers of 2',3'-dideoxycytidine 5'-triphosphate analogs as substrates for human DNA polymerases. Implications for the mechanism of toxicity.

5'-Triphosphates of beta-D and beta-L-enantiomers of 2',3'-dideoxycytidine (ddC), 2',3'-dideoxy-5-fluorocytidine (FddC), 1,3-dioxolane-cytidine (OddC), and 1,3-dioxolane-5-fluorocytidine (FOddC) were evaluated as inhibitors and substrates for human DNA polymerases alpha, beta, gamma, delta, and epsilon. L-ddCTP was not a substrate or inhibitor for any DNA polymerase studied; L-FddCTP was not an inhibitor or substrate for replicative DNA polymerases and was a less potent inhibitor of DNA polymerases gamma and beta than its D-enantiomer by 2 orders of magnitude. In contrast, all L-dioxolane analogs were potent inhibitors and chain terminators for all cellular DNA polymerases studied. The Ki values of their 5'-triphosphates for DNA polymerase gamma were found to be in the following order: D-ddC < D-FddC L-OddC D-FOddC < L-FOddC << L-FddC. The Ki values of L-OddCTP for the reactions catalyzed by DNA polymerases alpha, delta, epsilon, beta, and gamma were 6.0, 1.9, 0.4, 3.0, and 0.014 microM, respectively, and those of L-FOddCTP were 6.5, 1.9, 0.7, 19, and 0.06 microM, respectively. The Km values for incorporation of L-OddCTP into the standing points of primer extension were also evaluated and determined to be 1.3, 3.5, 1.5, 2.8, and 0.7 microM for DNA polymerases alpha, delta, epsilon, beta, and gamma, respectively. The incorporation of dioxolane analogs into DNA by replicative DNA polymerases could explain their potent cellular toxicity.

B virus and HIV in culture without much toxicity (10 -12). In contrast to L-SddC or L-FddC, L-OddC and L-FOddC were markedly more toxic than their D-enantiomers against human leukemic CEM cells (13). The spectrum of L-OddC toxicity against human tumor cells differed from that of cytosine arabinoside. L-OddC was also shown to have activity against solid human tumors in nude mice (14). Currently, L-OddC is being evaluated further as an anticancer agent. The reason why L-OddC but not L-ddC or L-SddC has potent antitumor activity is not clear, but it is probably associated with its interference in cellular DNA synthesis. L-SddC and L-OddC were shown to be phosphorylated by cellular kinases to the corresponding 5Ј-triphosphate metabolites inside cells. The antiviral effects of nucleoside analog are due to preferential interference with viral replication caused by incorporation into DNA by viral DNA polymerase. The interaction of 5Ј-triphosphate metabolites of L-OddC or L-FddC with cellular DNA polymerases may be one of the key factors in determining L-OddC cytotoxicity.
ssDNA cellulose was obtained from Sigma. DEAE cellulose (DE-52), phosphocellulose (P11), and S-Sepharose fast flow were purchased from Whatman. [␣-32 P]dCTP (3000 Ci/mmol), [␥-32 P]ATP (6000 Ci/mmol) and T4 polynucleotide kinase were obtained from Amersham Corp. M13 mp10 and M13 mp18 phage ssDNA were isolated as described previously (16). Oligonucleotides were synthesized on an Applied Biosystems 380A DNA synthesizer at the Yale Oligonucleotide Synthesis Facility. The primer oligonucleotides were labeled at the 5Ј position with T4 polynucleotide kinase using [␥-32 P]ATP, annealed to M13 phage ssDNA as described in Ref. 16, purified on a Sephadex G-25 column, and used as substrates for elongation reactions. BuAdATP was a kind gift from Professor George Wright (University of Massachussetts).
Enzymes-Human pol ␣, pol ␦, and pol ␥ from K562 chronic myelogenous leukemia cells and pol ␤ from KB cells were partially purified using P11, DE-52, S-Sepharose, and ssDNA cellulose chromatography (6). All DNA polymerases were distinguished from each other using several biochemical criteria, including chromatographic behavior, template preference, sensitivity to specific inhibitors and associated exonuclease activities. The ability to use poly(rA)⅐oligo(dT) 10 as template and primer in the presence of 80 mM KCl at pH 7.5-8.0 and high sensitivity to ddNTP indicated that the DNA polymerase activity was that of pol ␥ (17)(18)(19). pol ␦ and pol ␣ were eluted from a ssDNA-cellulose column at 0.32 M and 0.4 M KCl, respectively. Additional purifications were performed on DE-52 and S-Sepharose columns. pol ␦ and pol ␣ activities * This work was supported by NCI, National Institutes of Health, Grant CA44358 and National Institutes of Health Grant A13365. 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.
were monitored on poly(dA)⅐oligo(dT) [12][13][14][15][16][17][18] in the presence of proliferating cell nuclear antigen and activated DNA, respectively. pol ␣ was very sensitive to the inhibitor BuAdATP, displaying 25% residual activity in the presence of 5 M of this analog. In contrast, pol ␦ was much less sensitive, showing 80 -90% residual activity with the same concentration of drug. pol ␦ displayed significant 3Ј 3 5Ј exonuclease activity in contrast with the pol ␣ preparation, which showed only traces of the 3Ј 3 5Ј exonuclease activity, which were probably due to cross-contamination. As expected, the activity of pol ␦ was stimulated by proliferating cell nuclear antigen using poly(dA)⅐oligo(dT) 12-18 as template-primer, but proliferating cell nuclear antigen had no effect on pol ␣ (20 -22). pol ␤ that was separated from other polymerases using ssDNA cellulose chromatography was resistant to 3 mM N-ethylmaleimide (23). pol ⑀ was purified to near homogeneity from human placenta (24).
pol ␣ and pol ␦ activities were routinely assayed in 20 l of 20 mM Tris-HCl buffer (pH 7.4) containing 6 mM MgCl 2 , 1 mM dithiothreitol, 0.5 mM EDTA, 200 g/ml heat-inactivated bovine serum albumin, 150 g/ml activated calf thymus DNA, 20 M each of dATP, dTTP, and dGTP, and 1 Ci of [ 3 H]dCTP or [␣-32 P]dCTP at concentrations no less than the K m values for the appropriate enzymes. The same assay conditions were used for other polymerases with the following modifications: pol ␤, pH 8.5; pol ␥, pH 8.0 and 80 mM KCl; and pol ⑀, 8 mM MgCl 2 . Assays for DNA polymerases contained 2 l of enzyme (1 unit of pol ␣, pol ⑀, or pol ␥, 2 units of pol ␦, 0.75 unit of pol ␤). One unit of enzyme activity is defined as the amount of enzyme needed to incorporate 1 nmol of [ 3 H]dTMP/h into the acid-insoluble fraction at 37°C. The reactions were allowed to proceed at 37°C for 10 -60 min, after which 15-l aliquots were removed and spotted onto Whatman DE-81 discs. The filters were washed with 0.3 M KCl containing 0.5 mM EDTA and then fixed with ethanol. The incorporation of the radiolabeled substrate into the DNA chain was measured by liquid scintillation counting.
The K i values of each compound for all DNA polymerases studied were determined using a competitive inhibition equation as described in Ref. 25.
Chain Termination Assays-Triphosphates of ddCTP analogs were analyzed for their chain terminating activity using M13 phage DNA hybridized with a 5Ј-32 P-primer. Incorporation of one template-complementary nucleotide residue into the 3Ј-terminus of the primer was carried out in 8 l of a mixture containing the appropriate buffer solution as described above, enzyme, 0.01 M primer-template complex, and the analog under investigation. After incubation, the reaction was stopped by adding 4.5 l of formamide containing dyes and EDTA, and the reaction products were separated by denaturating 15% polyacrylamide gel electrophoresis. Chain termination assays in the presence of four natural dNTPs were based on previously published methods (26) and described in Fig. 6.
Quantitation of the Primer Extension Assay-The bands on the x-ray film, representing the incorporation of dNTP analogs into the standing position of 3Ј-ends of 5Ј-32 P-primers annealed with template, were quantitated with the aid of a densitometer (Molecular Dynamics) as described previously (27). The bands were scanned, and the efficiency of the primer extension and its dependence on substrate concentration were expressed as a Lineweaver-Burk plot (28). Experimental measurements were carried out within the linear range of the counts versus integration curve. The incubation time for all experiments was chosen to have a linear dependence between yield of product and time. The utilization of primer was usually less than 30%. Reaction times were 10 min for pol ␣ and pol ␦ and 15 min for pol ␤, pol ⑀, and pol ␥. K m values were estimated from these data.
Complexes of M13 phage DNA and primer used in this study were as follows.  Table I lists the K i values of L-and D-ddCTP analogs for these enzymes. As shown in Table I, L-OddCTP and L-FOddCTP were potent inhibitors of replicative DNA polymerases. The K i values of L-FOddCTP were about the same as those of its D-enantiomer for pol ␦, pol ␤, and pol ␥, but were less for pol ␣ and pol ⑀. The 5Ј-triphosphates of L-FddC and L-ddC at concentrations up to 20 M had no inhibitory effect on pol ␣, pol ␤, pol ␦, and pol ⑀. The results obtained for D-ddCTP and D-FddCTP were similar to those published previously (29,30). pol ␥ was much more sensitive than any other polymerase to all compounds tested with the exception of L-ddCTP. The triphosphate of L-OddC, which was shown to have anticancer activity (14), was a potent inhibitor of all DNA polymerase studied.
Chain Termination of DNA Synthesis by L-and D-Enantiomers of ddCTP Analogs-All analogs were evaluated for their chain-terminating activity in a system containing M13 phage DNA annealed with primer. Reaction products were monitored as described under "Experimental Procedures." The results for pol ␥, pol ␦, pol ⑀, and pol ␤ are illustrated in Figs. 2 and 3. pol ␥ could incorporate all analogs with the exception of L-ddCTP. pol ␦ (Fig. 2) and pol ⑀ (Fig. 3A) proved to be more selective enzymes with respect to their utilization of these analogs. Only L-OddCTP and L-FOddCTP were effective substrates for these enzymes. D-FOddCTP was 25 times less potent as a substrate for pol ⑀ than its L-enantiomer. Similar results were obtained for pol ␣. pol ␤ (Fig. 3B) incorporated D-ddCTP (lane 1), D-FddCTP (lanes 3 and 4), L-OddCTP (lanes 7 and 8), L-FOddCTP (lanes 9 and 10) and D-FOddCTP (lanes 11 and 12). L-ddCTP (lane 2) and L-FddCTP (lanes 5 and 6) were not substrates for pol ␤. The results obtained for D-ddCTP (lane 1) were similar to those previously described (23).
Kinetic Parameters for the Incorporation of L-OddCTP into a DNA Chain-The K m values for the incorporation of L-OddCTP into a growing DNA chain were determined (Fig. 4). Interaction of L-Nucleotide Analogs with Human DNA Polymerases no longer linear. Similar methods were applied for pol ␥, pol ⑀, and pol ␤. The K m values for dCTP and L-OddCTP are presented in Table II. In the evaluation of Table II, it should be emphasized that the kinetic values for the incorporation of natural nucleotides or analogs could be dependent on the nucleotide sequence of template-primer used (31). However, the ratio of the K m value of any nucleotide analog to K m value of the natural nucleotide should give a relative measure of the ability of that particular analog to be a substrate of DNA polymerase in comparison with the natural nucleotide.
As previously shown (6,20), L-SddCTP was a weak inhibitor of pol ␣ and pol ␦ and was not a substrate for either enzyme. We compared the ability of L-SddCTP and L-OddCTP to be incorporated at standing and running points of a DNA chain by pol ␣ under similar conditions. As one can see in Fig. 5, L-SddCTP is incorporated neither into the first nor into the eighth position of an elongated primer. Conversely, L-OddCTP is a good substrate for pol ␣, and the DNA fragments terminated by L-OddCMP are accumulated in both cases.
Chain Termination of DNA Synthesis by L-OddCTP in the Presence of All Four dNTPs-L-OddCTP was also assayed as a terminating substrate for pol ␣ in the presence of all four dNTPs following Sanger's method (26). The results are shown in Fig. 6. The comparison of the DNA sequence by Klenow fragment with D-ddCTP as a terminator are also presented. Synthesis was performed in the presence of M13 mp10 phage DNA annealed with 5Ј-32 P-15-mer primer. DNA reaction products were analyzed using 12% polyacrylamide gel electrophore-sis. The primer extension products of L-OddCTP were similar to the products observed for the reaction with D-ddCTP and Klenow fragment, as shown in lane 1 and lanes 2-5. These results indicate that L-OddCTP was incorporated into DNA at cytidine residue sites. Subsequent chase containing 100 M of four dNTPs (lane 4), which failed to extend the 3Ј-L-OddCMP terminated products, showed the ability of pol ␣ to incorporate L-OddCTP into the growing DNA chain in the presence of dCTP. An accumulation of DNA fragments was observed at a length expected for the chain terminator with cytidine as a base. DISCUSSION Recently, nucleoside derivatives with the unnatural L-configuration were evaluated against a broad spectrum of viruses, and some of these analogs proved to be very potent against  ␥ (lanes 1-9) and pol ␦ (lanes 10 -17). The reactions were performed as described under "Experimental Proce-  11 and 12, respectively). hepatitis B virus and HIV (6 -12). At the same time, differences in their cellular toxicities were noticed. In contrast to L-SddC or L-ddC, both L-and D-enantiomers of FOddC and OddC were shown to be very cytotoxic (13,14). The interaction of their 5Ј-triphosphate metabolites with cellular DNA polymerases may be one of the key factors in their varying toxicities. This possibility prompted us to evaluate the 5Ј-triphosphates of Land D-enantiomers of dioxolane-cytidine as potential inhibitors or substrates of human DNA polymerases. For comparison, nucleoside 5Ј-triphosphates with the natural D-configuration were included in the study. The results presented here show that L-OddCTP and its 5-fluoro-derivative are potent inhibitors and chain terminators of cellular DNA polymerases including pol ␣, pol ⑀, and pol ␦. The inhibition of polymerase activity is due to the incorporation of L-OddCTP into the DNA chain at cytidine residue sites but not to the inhibition of the rate of incorporation of other dNTPs. If the inhibition of synthesis was due to a dual mechanism including both incorporation and  ␣ (lanes 1-7) and pol ␦ (lanes 8 -14) into the 3-end of 5-32 P-15-mer primer annealed with M13 mp10 phage DNA. DNA primer extension was carried out as described under "Experimental Procedures" with the exception that incubation time was 10 min, and 0.2 unit of pol ␣ and 0.25 unit of pol ␦ were used to assure that less than 30% of the primer was consumed during experiments. A, incorporation of 2.5 uM dCTP (lanes 1 and 8); 0.2, 0.5, 1, 2, 5, and 10 uM L-OddCTP (lanes 2-7 and lanes 9 -14, respectively. B, the intensity of each track from A was quantitated by computer densitometry and plotted as 1/V versus 1/[S] for the reactions catalyzed by DNA pol ␣ (-ϩ-) and pol ␦ (-Ç-). 1/V is presented as relative value of density.  , lane 2, and B, lanes 1  and 2) and L-SddCTP (A, lanes 3 and 4, and B, lanes 3-5)  interference with the incorporation of other nucleosides, we would expect to see DNA fragments of a size inconsistent with DNA products terminated at cytidine residues (pauses) (Fig. 6). The interaction of L-OddCTP and L-FOddCTP with human DNA polymerases is the first example of the lack of enantioselectivity described for human replicative DNA polymerases. This phenomenon was described for HIV reverse transcriptase, pol ␥, and pol ␤ with respect to the L-and D-enantiomers of SddCTP (6,20,32) and to carbovir triphosphate (33). It should be mentioned that replicative DNA polymerases are believed to be more sensitive to changes in the conformation of substrates. Indeed, this is true for SddCTP whose L-isomer is not a substrate for replicative DNA polymerases. However, substitution of sulfur for oxygen at the 3Ј-position of the ribose residue leads to the opposite effect. L-OddCTP is a good terminating substrate for replicative DNA polymerases, inhibiting them with K i values in the pharmacologically relevant concentration range.

FIG. 2. Incorporation of L-and D-enantiomers of ddCTP analogs into the 3-end of a 17-mer primer annealed with M13 mp18 phage DNA by pol
The amount of L-OddCMP present at the DNA terminus depends not only on the efficiency of incorporation of L-OddCTP by polymerases but also on the rate of excision by 3Ј 3 5Јexonucleases (34). The ability of 3Ј 3 5Ј-exonucleases associated with pol ␦ and pol ⑀ are currently evaluated. The anticancer activity of L-OddC may be a result of termination of DNA synthesis after L-OddCTP incorporation into proliferating cells coupled with inefficient excision of incorporated L-OddCMP from DNA.
In the present study, we also made an attempt to address the impact of the substitution of fluorine for hydrogen at the 5-position of cytidine 5Ј-triphosphate analogs on their ability to serve as substrates for cellular DNA polymerases. As shown in Table I, both L-and D-enantiomers of ddCTP and FddCTP were not substrates for pol ␣, pol ⑀, and pol ␦ and inhibited pol ␤ and pol ␥ at the same range of concentrations. Similar results were obtained for the L-enantiomers of OddCTP and FOddCTP. No significant differences were seen between L-OddCTP and its 5-fluoro-derivatives in terms of their interaction with DNA polymerases. Both L-enantiomers were equally potent inhibitors for these DNA polymerases. With respect to pol ␤, L-OddCTP was 6 times more potent than its 5-fluoro-analog. At present, we cannot explain the differences in the interaction of L-OddCTP and its 5-fluoro-derivative with pol ␤.
In summary, L-OddC is the first L-nucleoside shown to have potent antitumor activity. This activity could be related to the ability of L0OddCTP to be utilized as a substrate by human replicative DNA polymerases. Surprisingly, this property is not shared with L-SddC and L-ddC, which are relatively noncytotoxic. The discovery that the L-enantiomers of chain-terminating nucleotides can be incorporated by replicative DNA polymerases could lead to the development of a new class of anticancer compounds.