Stereoisomers of deoxynucleoside 5'-triphosphates as substrates for template-dependent and -independent DNA polymerases.

All four possible stereoisomers of dNTP with regard to deoxyribofuranose C-1' and C-4' carbon atoms were studied as substrates for several template-dependent DNA polymerases and template-independent terminal deoxynucleotidyl transferase. It was shown that DNA polymerases alpha, beta, and epsilon from human placenta and reverse transcriptases of human immunodeficiency virus and avian myeloblastosis virus incorporate into the DNA chain only natural beta-D-dNTPs, whereas calf thymus terminal deoxynucleotidyl transferase incorporates two nucleotide residues of alpha-D-dNTP and extends the resulting oligonucleotide in the presence of beta-D-dNTPs. The latter enzyme also extended alpha-anomeric D-oligodeoxynucleotide primers in the presence of beta-D-dNTPs. None of the studied enzymes utilized L-dNTPs. These data indicate that template-dependent DNA polymerases are highly stereospecific with regard to dNTPs, whereas template-independent terminal deoxynucleotidyl transferase shows less stereodifferentiation. It is likely that the active center of the latter enzyme forms no specific contacts with the nucleic bases of both nucleotide substrate and oligonucleotide primer.

The substrate activity of the stereoisomers of DNA polymerase natural substrates and their analogs is of significant current interest, because these compounds can help to ascertain the mechanism of substrate binding by DNA polymerases and identify the parts of the dNTP molecule that specifically bind to the active center of the enzymes. Indeed, the stereoisomers of natural ␤-D-dNTPs 1 differ only in the mutual orientation of the reaction center (triphosphate fragment), the genetic recognition element (nucleic base), and the sugar residue.
To date, no compounds have been found that would specifically inhibit TDT in cell cultures. One of the reasons for that is the similarity in substrate specificity between TDT and some other DNA polymerases, especially DNA polymerase ␤ (11,12) and endogeneous reverse transcriptases (13).
However, the independence of TDT from the template is a factor that could simplify the design of selective inhibitors of this enzyme. Indeed, we have recently found (4) that dNTP analogs with trans-like orientation of the nucleic base and triphosphate residue efficiently and selectively inhibit DNA synthesis catalyzed by TDT.
In this paper we synthesized all four possible stereoisomers of dNTPs with respect to the C-1Ј and C-4Ј carbon atoms and evaluated them as substrates for template-dependent DNA polymerases ␣, ␤, and ⑀ from human placenta, reverse transcriptases from HIV and avian myeloblastosis virus, and TDT from calf thymus. We also synthesized two anomeric ␣-D-oligodeoxynucleotides and studied them as primers for TDT. * This work was supported by Russian Foundation for Basic Research Grants N93-04-20542 and N95-03-08142a and the French CNRS (project "Cooperation Franco-russe" 1752). 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.
Enzymes and DNA-HIV reverse transcriptase was isolated according to Ref. 19. DNA polymerases ␣ and ⑀ were isolated from human placenta as described in Ref. 20; DNA polymerase ␤ was purified according to Ref. 21. Avian myeloblastosis virus reverse transcriptase and calf thymus TDT were from Omutninsk Chemicals (Russia) and Amersham Corp., respectively.
Single-stranded M13mp10 DNA was isolated from the culture medium of the recipient E. coli K12XL1 strain as described in Ref. Primer Extension Assays-For the template-dependent DNA polymerases, the assay mixture (volume, 6 l) contained 0.01 M templateprimer (Fig. 2), stereoisomer under study or its natural counterpart, enzyme (2 activity units of reverse transcriptase or 1 unit of DNA polymerases ␣ and ␤), and the corresponding buffer. The reaction was carrried out for 20 min at 37°C and terminated by adding 3 l of deionized formamide containing 0.5 mM EDTA and 2% bromphenol blue and xylene cyanol. The reaction products were separated by electrophoresis in 20% polyacrylamide gel, and the gels obtained were autoradiographed.

RESULTS
The structure of the dNTP stereoisomers studied is shown in Fig. 1.
The fidelity of DNA synthesis catalyzed by HIV reverse transcriptase is rather low (24 -26), and the probability of incorrect dNMP incorporation is higher for the template positions remote from the primer 3Ј terminus by more than two nucleotide residues. Therefore, we used different template-primers for the dCTP, dTTP, and dATP stereoisomers (type of template-primer used is specified in figure caption).
It is evident from Fig. 3 that ␤-L-dCTP is not incorporated into the DNA chain by HIV reverse transcriptase (Fig. 3A,  lanes 4 -6). In the control assays, dATP (Fig. 3A, lane 2) and dATP ϩ dCTP (Fig. 3A, lane 3) were used. Similar results were obtained for avian myeloblastosis virus reverse transcriptase (data not shown). We also evaluated ␤-L-dCTP and ␤-L-dTTP as substrates for human DNA polymerases ␣, ␤, and ⑀. It can be seen in Fig. 4 that ␤-L-dCTP is also not incorporated into the DNA chain by these enzymes (Fig. 4, A and B, lanes 4 -6). It was not a substrate for DNA polymerase ⑀ (data not shown). Fig. 3B presents the results of TDT assays in the presence of ␤-L-dCTP. Clearly, some incorporation is observed at 2 (Fig. 3B, lane 5) and 20 (Fig. 3B, lane 6) M, but scanning densitometry revealed that the extent of primer conversion is only 2% at 20 M. The ␤-L-dTTP displayed the same substrate activity as ␤-L-dCTP for all DNA polymerases studied (data not shown). It can be seen in Fig. 5 that ␣-L-dNTPs (lanes 3, 4, 6, and 7) and ␣-D-dNTPs (lanes 8, 9, 11, and 12) are not substrates for HIV reverse transcriptase. We assume that ␣-L-dATP practically does not interact with the DNA-synthesizing complex, because it did not inhibit primer extension by dTMP and dGMP residues even at a high concentration (Fig. 5, lanes 6 -7), whereas ␣-D-dATP completely inhibited primer extension at a [␣-D-dATP]/[dNTP] concentration ratio of 10:1 (Fig. 5, lane 12). We attribute the presence of a weak heptadecanucleotide band in lane 2 of Fig. 5 to the error-prone properties of HIV reverse transcriptase. Both ␣-D-dNTPs and ␣-L-dNTPs were not utilized as substrates by human DNA polymerases (data not shown).
It is evident from Fig. 6 that two ␣-D-dTMP (lanes 10 -12) and ␣-D-dAMP (lanes 13-15) residues are incorporated into the primer by TDT, and one more residue is incorporated less efficiently. The efficiency of primer extension depended on the substrate concentration. Similar results were obtained for ␣-D-dCTP (data not shown). The ␣-L-dTTP (Fig. 6, lanes 4 -6) and ␣-L-dATP were not utilized by the enzyme. In the control assays (Fig. 6, lanes 2 and 3), dTTP was used as a substrate. We ascribe the presence of weak pentadecanucleotide bands on lanes 1, 5, and 6 of Fig. 6 to contamination of the TDT preparation with trace amounts of dNTPs.
Then, we examined the ability of TDT to elongate oligonucleotides terminated by ␣-D-dNMP residues in the presence of different concentrations of dNTPs. It can be seen in Fig. 7 that oligodeoxynucleotides containing ␣-D-dTMP and ␣-D-dAMP residues at the 3Ј end are elongated by TDT in the presence of 0.2, 2, and 20 M dNTPs (lanes 6 -8 and 11-13), the length of the products being dependent on the dNTP concentration.
At the second stage of this research we evaluated ␣-d[(Tp) 11 T] and ␣-d[(Tp) 3 T] as primers for TDT in the presence of natural dNTPs and ␣-dNTP (Fig. 8, B and D). Oligonucleotides ␤-d[(Tp) 9 T] (Fig. 8A) and ␤-d[(Tp) 3 T] (Fig. 8C) were used as reference primers. All four oligothymidylates were extended in the presence of dGTP, and the length of the products depended on the dGTP concentration (Fig. 8, lanes 2-4). Similar results were obtained with dCTP as substrate. DISCUSSION Our results indicate that template-dependent DNA polymerases utilize as substrates only ␤-D-dNTPs, whereas for template-independent TDT ␣-D-dNTPs but not ␤-L-dNTPs or ␣-L-dNTPs are substrates. It is noteworthy that ␣-D-dNTPs but not ␤-L-dNTPs or ␣-L-dNTPs inhibited primer extension catalyzed by HIV and avian myeloblastosis virus reverse transcriptases. It seems likely that the 3Ј hydroxyl of L-dNTPs hinders formation of the productive [DNA polymerase ϩ template-primer ϩ dNTP] complex by creating steric obstacles for dNTP binding to the enzyme.
The inconsistency between our results and the data of Focher et al. (2) may be attributed to the following differences in the experimental conditions. First, these authors used a homopolymeric template-primer, whereas in our experiments random-   20 , because these enzymes do not possess 3Ј 3 5Ј exonuclease activity. Indeed, ion-exchange chromatography performed in Ref. 2 cannot properly separate inorganic pyrophosphate from ␤-L-dTTP, because these compounds have close charges under the conditions described. Thus, it is possible that pyrophosphorolysis led to formation of ␤-D-dTTP in the assay mixture and subsequent primer extension. Therefore, for our part, we additionally purified ␤-L-dTTP and ␤-L-dCTP by reversed-phase HPLC.
Interestingly, the affinity of 3Ј-modified ␤-L-dNTP analogs to human DNA polymerases ␣, ␤, ⑀, and ␥ and HIV reverse transcriptase drops as the bulk of the substituent is increased, O Ͼ S Ͼ CH Ϸ CH 2 Ͼ CH 2 OH (Refs. 7 and 10; this paper for human DNA polymerases; Refs. 3, 7, and 9 for HIV reverse transcriptase). It is possible that the dNTP-binding site of DNA polymerases contains one or several groups near the C-3Ј atom of ␤-L-dNTP, which hinders productive dNTP binding.
It has earlier been shown (4) that carbocyclic ␣-Dand L-dNTP analog isosteres I and II (Fig. 9) are incorporated into the DNA chain by TDT, but are not utilized by template-dependent DNA polymerases. In this work we showed that ␣-D-dNTPs are substrates for TDT but are not recognized by template-dependent DNA polymerases, suggesting that they could be used as selective inhibitors of TDT to elucidate the functional role of this enzyme. On the other hand, unlike their carbocyclic counterpart II, ␣-L-dNTPs were not incorporated into the primer by TDT. We ascribe this difference in substrate properties to the presence of a double C-2Ј-C-3Ј bond in II. Indeed, the latter is known to impart planarity to the sugar moiety and drastically increase the affinity of the dNTP to DNA polymerases (4,27,28). It is likely that ␣-L-dNTPs do not interact productively with TDT because of the difference in the position of the 3Јhydroxyl in L-and D-dNTPs.
Because ␤-oligodeoxyribonucleotides containing one or two ␣-D-dNMP residues at the 3Ј terminus are extended by TDT in the presence of natural dNTPs (Fig. 7), it seemed interesting to evaluate oligodeoxyribonucleotides containing only ␣-D-dNMP residues as a primer for TDT. Another goal at this stage of the research was to find oligonucleotides utilized as primers by TDT and stable in the cell and blood, i.e. resistant toward nucleases. Short ␣-oligonucleotides are interesting in this respect due to their increased stability in serum (29). In this work we compared two ␣-oligothymidylates with their ␤ counterparts as primers for TDT and found that all four oligonucleotides are extended by this enzyme, although the ␣-oligothymidylates are slightly less efficient as primers (Fig. 8).
The results obtained may imply that the nucleic bases of dNTP and oligonucleotide primers do not bind in a specific manner to the substrate-and primer-binding sites, respectively, of the TDT active center. Indeed, we have found that phenylphosphonyldiphosphate (III, Fig. 9) was not a substrate for TDT (data not shown), but inhibited TDT-catalyzed primer extension by 50% at a [III]/[dTTP] molar concentration ratio of 1:1.