Modified Nucleotides as Substrates of Terminal Deoxynucleotidyl Transferase

The synthesis of novel modified nucleotides and their incorporation into DNA sequences opens many possibilities to change the chemical properties of oligonucleotides (ONs), and, therefore, broaden the field of practical applications of modified DNA. The chemical synthesis of nucleotide derivatives, including ones bearing thio-, hydrazino-, cyano- and carboxy groups as well as 2-pyridone nucleobase-containing nucleotides was carried out. The prepared compounds were tested as substrates of terminal deoxynucleotidyl transferase (TdT). The nucleotides containing N4-aminocytosine, 4-thiouracil as well as 2-pyridone, 4-chloro- and 4-bromo-2-pyridone as a nucleobase were accepted by TdT, thus allowing enzymatic synthesis of 3’-terminally modified ONs. The successful UV-induced cross-linking of 4-thiouracil-containing ONs to TdT was carried out. Enzymatic post-synthetic 3’-modification of ONs with various photo- and chemically-reactive groups opens novel possibilities for future applications, especially in analysis of the mechanisms of polymerases and the development of photo-labels, sensors, and self-assembling structures.


Incorporation of Modified Nucleotides by Terminal Deoxynucleotidyl Transferase
In order to determine whether these novel modified nucleotides can be incorporated into ON, primer extension reactions (PEX) were conducted using TdT. Among all DNA polymerases TdT is unique for its broad utilization of divalent metal ions. TdT is able to use several metal cations such as Mg 2+ , Zn 2+ , Co 2+ , Mn 2+ [65]. Furthermore, each divalent metal ion contributes to the kinetics of nucleotide incorporation differently. It has been shown that in the presence of Co 2+ TdT preferentially incorporates purine nucleotides while Mg 2+ facilitates incorporation of pyrimidine nucleotides [66]. Zn 2+ can also improve template-independent synthesis [67]. To monitor the discrimination between incorporation of modified nucleotides by TdT two buffer systems were used. In our studies the first system is based on sodium glutamate and Mg 2+ which is likely to be the basic enzyme cofactor in This method allowed accessing the triphosphate from the appropriate modified nucleoside in 20% yield. The triphosphates were best isolated by ion exchange chromatography using a diethylaminoethyl (DEAE) Sephadex A-25 column with a gradient of LiCl, followed by precipitation of modified nucleotides from acetone/methanol mixtures. The one step phosphorylation procedure allowed us to prepare the nucleoside triphosphates with >95% purity in a fast way for further studies.

Incorporation of Modified Nucleotides by Terminal Deoxynucleotidyl Transferase
In order to determine whether these novel modified nucleotides can be incorporated into ON, primer extension reactions (PEX) were conducted using TdT. Among all DNA polymerases TdT is unique for its broad utilization of divalent metal ions. TdT is able to use several metal cations such as Mg 2+ , Zn 2+ , Co 2+ , Mn 2+ [65]. Furthermore, each divalent metal ion contributes to the kinetics of nucleotide incorporation differently. It has been shown that in the presence of Co 2+ TdT preferentially incorporates purine nucleotides while Mg 2+ facilitates incorporation of pyrimidine nucleotides [66].
Zn 2+ can also improve template-independent synthesis [67]. To monitor the discrimination between incorporation of modified nucleotides by TdT two buffer systems were used. In our studies the first system is based on sodium glutamate and Mg 2+ which is likely to be the basic enzyme cofactor in vivo. Contrary, another buffer system which is referred to as optimal buffer for TdT is absolutely artificial compared to in vivo conditions as it contains potassium cacodylate and Co 2+ . Moreover thinking of Co 2+ as a versatile metal ion cofactor capable of playing a role of both Mg 2+ and Zn 2+ , a more efficient nucleotide incorporation using latter buffer system could be anticipated [68]. Figures 1 and 2 illustrate non-templated 3'-end elongation using 2-pyridone-based nucleoside triphosphates and compounds 3 (4-thio-2'-deoxyuridine triphosphate, 4-thio-dUTP), 4 (N 4 -amino-2'deoxycytidine triphosphate, N 4 -amino-dCTP), 7 (5-cyano-2'-deoxyuridine triphosphate, dU 5CN TP), 8 (5-carboxy-2'-deoxyuridine triphosphate, dU 5COOH TP) catalyzed by TdT in Mg 2+ -containing and its optimal (Co 2+ -containing) buffer system, respectively. In general, the utilization of the optimal buffer for TdT resulted in more efficient incorporation of natural nucleotides as well as non-natural analogues ( Figure 2). It should be noted that in the case of optimal reaction conditions TdT utilizes nearly every member of our library of non-natural nucleotides presented here whereas more than a half of compounds tested were unsuitable substrates for TdT under milder (glutamate/Mg 2+ ) conditions. This in turn proves already established strict dependency of TdT catalytic activity and more importantly substrate specificity on reaction buffer.
It was previously demonstrated that 2-pyridone-based nucleotides are utilized by several template-dependent DNA polymerases [69]; however mostly inefficiently since only a single nucleotide was incorporated with no further elongation. Data provided in Figure 1 (lanes 7 and 8) revealed that longer products (up to ten extra nucleotides) were generated using 4-chloro-2pyridone-2'-deoxyriboside triphosphate (dPyr 4Cl TP) and 4-bromo-2-pyridone-2'-deoxyriboside triphosphate (dPyr 4Br TP) as substrates than with equivalent concentrations of dUTP and dCTP suggesting that these compounds were better substrates for TdT. In addition, a substantial increase in the length of the elongation products by these two compounds in optimal environment was observed (Figure 2, lanes 7 and 8). Hence, in order to introduce varying number of monomers for the template-independent synthesis of halogenated nucleic acids, different buffers can be applied. vivo. Contrary, another buffer system which is referred to as optimal buffer for TdT is absolutely artificial compared to in vivo conditions as it contains potassium cacodylate and Co 2+ . Moreover thinking of Co 2+ as a versatile metal ion cofactor capable of playing a role of both Mg 2+ and Zn 2+ , a more efficient nucleotide incorporation using latter buffer system could be anticipated [68]. Figures 1 and 2 illustrate non-templated 3'-end elongation using 2-pyridone-based nucleoside triphosphates and compounds 3 (4-thio-2'-deoxyuridine triphosphate, 4-thio-dUTP), 4 (N 4 -amino-2'deoxycytidine triphosphate, N 4 -amino-dCTP), 7 (5-cyano-2'-deoxyuridine triphosphate, dU 5CN TP), 8 (5-carboxy-2'-deoxyuridine triphosphate, dU 5COOH TP) catalyzed by TdT in Mg 2+ -containing and its optimal (Co 2+ -containing) buffer system, respectively. In general, the utilization of the optimal buffer for TdT resulted in more efficient incorporation of natural nucleotides as well as non-natural analogues ( Figure 2). It should be noted that in the case of optimal reaction conditions TdT utilizes nearly every member of our library of non-natural nucleotides presented here whereas more than a half of compounds tested were unsuitable substrates for TdT under milder (glutamate/Mg 2+ ) conditions. This in turn proves already established strict dependency of TdT catalytic activity and more importantly substrate specificity on reaction buffer.
It was previously demonstrated that 2-pyridone-based nucleotides are utilized by several template-dependent DNA polymerases [69]; however mostly inefficiently since only a single nucleotide was incorporated with no further elongation. Data provided in Figure 1 (lanes 7 and 8) revealed that longer products (up to ten extra nucleotides) were generated using 4-chloro-2pyridone-2'-deoxyriboside triphosphate (dPyr 4Cl TP) and 4-bromo-2-pyridone-2'-deoxyriboside triphosphate (dPyr 4Br TP) as substrates than with equivalent concentrations of dUTP and dCTP suggesting that these compounds were better substrates for TdT. In addition, a substantial increase in the length of the elongation products by these two compounds in optimal environment was observed (Figure 2, lanes 7 and 8). Hence, in order to introduce varying number of monomers for the template-independent synthesis of halogenated nucleic acids, different buffers can be applied. To the contrast, severe discrimination between two buffers was observed in the case of 2-pyridone-2'-deoxyriboside triphosphate (dPyrTP) and 4-hydroxy-2-pyridone-2'-deoxyriboside triphosphate (dPyr 4OH TP) acting as substrates for TdT. Figure 1 illustrates that utilization of both dPyrTP and dPyr 4OH TP by TdT was not detectable in glutamate/Mg 2+ buffer (lanes 5 and 6). However in the presence of Co 2+ TdT uses the former as substrate for the incorporation and subsequent To the contrast, severe discrimination between two buffers was observed in the case of 2-pyridone-2'-deoxyriboside triphosphate (dPyrTP) and 4-hydroxy-2-pyridone-2'-deoxyriboside triphosphate (dPyr 4OH TP) acting as substrates for TdT. Figure 1 illustrates that utilization of both dPyrTP and dPyr 4OH TP by TdT was not detectable in glutamate/Mg 2+ buffer (lanes 5 and 6). However in the presence of Co 2+ TdT uses the former as substrate for the incorporation and subsequent elongation while only a single incorporation of dPyr 4OH is observed (Figure 2, lanes 5 and 6). Recently, it was demonstrated that binding affinity of TdT to the 3'-end of ON modified with 3-hydroxy-4-pyridone-based nucleotides depended on the Mg 2+ concentration and caused formation of unfavourable secondary structure of 3'-terminus [70]. As a consequence after incorporation of single nucleotide further elongation was impeded. Our results support such interpretation and further suggest that not only buffer composition but also additional nucleobase modifications may lead to steric hindrance. elongation while only a single incorporation of dPyr 4OH is observed (Figure 2, lanes 5 and 6). Recently, it was demonstrated that binding affinity of TdT to the 3'-end of ON modified with 3-hydroxy-4pyridone-based nucleotides depended on the Mg 2+ concentration and caused formation of unfavourable secondary structure of 3'-terminus [70]. As a consequence after incorporation of single nucleotide further elongation was impeded. Our results support such interpretation and further suggest that not only buffer composition but also additional nucleobase modifications may lead to steric hindrance. 4-Thiouracil nucleobase is naturally occurring in some transfer RNA species and is widely used as a photoactivatable RNA-RNA, RNA-DNA, RNA-protein cross-linking agent [71]. Hence, T7 RNA polymerase uses 4-thio-UTP as a substrate during transcription in vitro, while it is known that 4-thiothymidine is readily incorporated into DNA during replication [45,46]. The data presented here revealed that 4-thio-dU was incorporated into ON with similar efficiency to its natural counterpart dU possibly due to its very similar structure. Despite the fact that Co 2+ -containing buffer system enhances utilization of certain nucleotide analogues, only a slight increase in incorporation of 4-thio-dU compared to glutamate/Mg 2+ buffer was observed ( Figure 2, lane 10). Nevertheless, these results suggest 4-thio-dUTP as a promising photo-affinity label for DNA. N 4 -Amino-dCTP appeared to be a moderate substrate for TdT in both buffer systems compared to native nucleoside triphosphates. On the other hand, an ON bearing at least several hydrazine modifications could be further modified or cross-linked to nucleic acid molecule, protein or appropriately pre-treated surface.
Overall, dPyrTP, dPyr 4Cl TP, dPyr 4Br TP, 4-thio-dUTP and N 4 -amino-dCTP emerged as substrates used by TdT for incorporation and subsequent elongation whereas dPyr 4OH and dU 5CN were only incorporated but further elongation was halted. Only a single incorporation of dU 5CN is slightly confusing as it is known that TdT is able to incorporate modified pyrimidine nucleotides with bulky groups at fifth position [30,72]. Neither dPyr 5COOH TP nor dU 5COOH TP were used by TdT as substrates under both conditions tested. It might be speculated that a negative charge of both nucleotides was detrimental to an activity of TdT. 4-Thiouracil nucleobase is naturally occurring in some transfer RNA species and is widely used as a photoactivatable RNA-RNA, RNA-DNA, RNA-protein cross-linking agent [71]. Hence, T7 RNA polymerase uses 4-thio-UTP as a substrate during transcription in vitro, while it is known that 4-thiothymidine is readily incorporated into DNA during replication [45,46]. The data presented here revealed that 4-thio-dU was incorporated into ON with similar efficiency to its natural counterpart dU possibly due to its very similar structure. Despite the fact that Co 2+ -containing buffer system enhances utilization of certain nucleotide analogues, only a slight increase in incorporation of 4-thio-dU compared to glutamate/Mg 2+ buffer was observed (Figure 2, lane 10). Nevertheless, these results suggest 4-thio-dUTP as a promising photo-affinity label for DNA. N 4 -Amino-dCTP appeared to be a moderate substrate for TdT in both buffer systems compared to native nucleoside triphosphates. On the other hand, an ON bearing at least several hydrazine modifications could be further modified or cross-linked to nucleic acid molecule, protein or appropriately pre-treated surface.
Overall, dPyrTP, dPyr 4Cl TP, dPyr 4Br TP, 4-thio-dUTP and N 4 -amino-dCTP emerged as substrates used by TdT for incorporation and subsequent elongation whereas dPyr 4OH and dU 5CN were only incorporated but further elongation was halted. Only a single incorporation of dU 5CN is slightly confusing as it is known that TdT is able to incorporate modified pyrimidine nucleotides with bulky groups at fifth position [30,72]. Neither dPyr 5COOH TP nor dU 5COOH TP were used by TdT as substrates under both conditions tested. It might be speculated that a negative charge of both nucleotides was detrimental to an activity of TdT.
So far, no atoms have been unambiguously identified between the amino acid residues of TdT and the primer nucleobases that reside in sufficient proximity to one another to be considered a polar interaction, thus indicating that binding must rely entirely upon interaction with the sugar-phosphate backbone. It is assumed that the broad substrate specificity of TdT is based on polymerase insensitivity to the chemical structure of an incoming dNTP although the structure and accessibility of the 3'-terminus of ONs is crucial [26,28]. In addition, it has been proposed that kinetics of extension does not appear to be influenced by π-electron surface area of nucleotide analogues but rather by the size of modified nucleotides [11]. For example, it was shown that smaller nucleotides bearing 5-fluoroindolyl-and 5-nitroindolyl-modifications were more efficiently elongated whereas their bulkier counterparts (e.g., 5-napthylindolyl-modified), have been refractory to elongation. Hence, the inability to elongate large, bulky, non-natural nucleotides likely results from steric constraints [11,73]. However, the clear difference between efficiency of incorporation of dPyr 4OH , dPyr 4Cl (or dPyr 4Br ) and 4-thio-dU as well as poor elongation of 5-cyano-modified dU by TdT cannot be explained in this way. TdT requires at least three deoxynucleotide residues on the primer strand for an efficient catalysis of the tailing reaction [31]. Subsequently, when polymerase reaches the extended section, 3'-end modifications could block the entrance of an incoming dNTP by enhancing primer affinity to TdT. Consequently reaction may stall explaining why only limited number of residues are appended, e.g., in the case of 4-thio-dU. The same result would also be observed if the 3'-modified primer fails to interact with the active center of TdT or it folds into unfavourable secondary structures as in the case of pyrene-2'-deoxynucleotide [26] or 2-methyl-3-hydroxypyridin-4(1H)-one 2'-deoxynucleotide (dH) [74]. In our hands TdT does not extend the 4-thio-dU-elongated primer in the presence of deoxythymidine triphosphate (dTTP) (data not shown); a similar effect has been observed with dH [70].

UV Cross-Linking of Oligonucleotides to Terminal Deoxynucleotidyl Transferase
In order to examine the potential of thio-modified ON to serve as a cross-linking agent, UVA-induced linking assay was conducted. Successful PEX using both 4-thio-dUTP and dUTP (a control reaction) were carried out where an average of 6 and 15 nucleotides were incorporated, respectively (data not shown). The next step was to select the model protein of interest to be cross-linked to 4-thiouridine-containing ON (4-thio-dU-ON). It was decided to cross-link the modified ON to TdT for several reasons. First of all, it is obvious that using TdT eliminates the need to search for proteins which specifically interact with the 3'-end of modified ONs. Secondly, it could be anticipated that, after catalysis, thio-groups would be located adjacent to amino acid side chains in the active site of TdT and thus revealing an actual interaction with nucleobase(s). Due to the fact that the generation of photo-induced complexes strongly depends on the distance between a photo-reactive group and amino acid residues it has been suggested that a successful cross-linking would be an evidence, even if indirect, of interactions between nucleobases and TdT. Figure 3 shows the effect of exposing PEX mixtures to 365 nm UV light. Based on the slower mobility of the 33 P label it was clear that the 4-thio-dU-ONs cross-linked to TdT (Figure 3, lanes 4-6). Data provided in Figure 3 (lanes 4 and 5) indicate that induction of cross-links depended upon a dose of UVA irradiation. A vast excess of TdT over ON did not improve cross-linking efficiency (Figure 3, lane 6). Moreover, it was elucidated that no cross-linked complexes were detected during irradiation at lower intensity (data not shown). The free TdT has a molecular mass of 45-47 kDa as determined by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE), whereas cross-linked complex run at the position of 55 kDa marker (Figure 3, lane 1 and lanes 4-6, respectively). This suggests a 1:1 covalent complex between TdT and 4-thio-dU-ON (molecular mass of 8-10 kDa). A second complex could be noted with a molecular mass of ~130 kDa though exact constitution of this heavier complex could not be ascertained easily (Figure 3, lane 4). It might be speculated that formation of more complicated crosslinks was inevitable since ON could contain up to ten 4-thiouracil nucleobases, which theoretically might be triggered to generate at least several cross-links, whether it would be covalent bonds between two ONs, two protein molecules or aggregates of mixed origin. No cross-linking was observed in the absence of UV light and in the presence of dU-ONs.
TdT has a pivotal role in vivo to generate junctional diversity during V(D)J recombination by adding random nucleotides [20]. A detailed mechanism of antibody gene recombination still needs to be puzzled out. Although the biological role of TdT is tightly connected to its ability to utilize a wide variety of substrates in vitro and in vivo, there is no exact explanation on substrate selectivity of TdT up to date [75]. Data presented here slightly contradicts with the already established assumption that a broad substrate selectivity of TdT is based on unspecific interaction with sugarphosphate backbone rather than specific contacts with nucleobases. Formation of UV-induced complexes between 4-thio-dU-ONs and TdT suggests an existing juxtaposition of the 4-thiouracil base with the appropriate amino acid side chains in the TdT.
All these observations clearly show that multiple interactions between the substrate and TdT take place and the nature of nucleobase plays an important role with respect to nucleotide selection and chain elongation. Therefore, further studies are needed to elucidate a substrate selectivity of TdT. The free TdT has a molecular mass of 45-47 kDa as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), whereas cross-linked complex run at the position of 55 kDa marker (Figure 3, lane 1 and lanes 4-6, respectively). This suggests a 1:1 covalent complex between TdT and 4-thio-dU-ON (molecular mass of 8-10 kDa). A second complex could be noted with a molecular mass of~130 kDa though exact constitution of this heavier complex could not be ascertained easily (Figure 3, lane 4). It might be speculated that formation of more complicated cross-links was inevitable since ON could contain up to ten 4-thiouracil nucleobases, which theoretically might be triggered to generate at least several cross-links, whether it would be covalent bonds between two ONs, two protein molecules or aggregates of mixed origin. No cross-linking was observed in the absence of UV light and in the presence of dU-ONs.

General Information
TdT has a pivotal role in vivo to generate junctional diversity during V(D)J recombination by adding random nucleotides [20]. A detailed mechanism of antibody gene recombination still needs to be puzzled out. Although the biological role of TdT is tightly connected to its ability to utilize a wide variety of substrates in vitro and in vivo, there is no exact explanation on substrate selectivity of TdT up to date [75]. Data presented here slightly contradicts with the already established assumption that a broad substrate selectivity of TdT is based on unspecific interaction with sugar-phosphate backbone rather than specific contacts with nucleobases. Formation of UV-induced complexes between 4-thio-dU-ONs and TdT suggests an existing juxtaposition of the 4-thiouracil base with the appropriate amino acid side chains in the TdT.
All these observations clearly show that multiple interactions between the substrate and TdT take place and the nature of nucleobase plays an important role with respect to nucleotide selection and chain elongation. Therefore, further studies are needed to elucidate a substrate selectivity of TdT.

General Information
Chemicals and solvents were purchased from Sigma-Aldrich (Steinheim, Germany) and Alfa Aesar (Karlsruhe, Germany) and used without further purification. Thin-layer chromatography . Chemical shifts (δ) are reported in ppm relative to the solvent resonance signal as an internal standard. UV spectra were recorded on a Lambda 25 UV/VIS spectrometer (Perkin Elmer, Singapore). High-performance liquid chromatography mass-spectrometry (HPLC-MS) analyses were performed using a high performance liquid chromatography system, equipped with a photo diode array detector (SPD-M20A) and a mass spectrometer (LCMS-2020, Shimadzu, Kyoto, Japan) equipped with an electrospray ionization (ESI) source. The chromatographic separation was conducted using a YMC Pack Pro column, 3 × 150 mm (YMC, Kyoto, Japan) at 40 • C and a mobile phase that consisted of 0.1% formic acid water solution or 5 mM ammonium acetate buffer (solvent A), and acetonitrile (solvent B). Mass spectrometry data was acquired in both positive and negative ionization mode and analyzed using the LabSolutions LCMS software (Software version 5.42 SP6, Shimadzu, Kyoto, Japan). (iii) 4-Thio-2'-deoxyuridine (2). A mixture of 3',5'-bis-O-acetyl-4-thio-2'-deoxyuridine (840 mg, 2.55 mmol) and 1M sodium methylate (5.1 mL) was stirred for 30 min at room temperature. The reaction was monitored with TLC. The crude reaction mixture was purified by reverse phase column chromatography (C-18 cartridges, water/methanol mixture, 10:0→10:2). The solvents were removed under reduced pressure to afford yellowish solid reaction product, mp 147-150

Formation of 4-Thio-dU-ON:Protein Cross-links
The cross-linking apparatus was constructed with slight modifications as described previously [77]. It consisted of an ice container, 96-well plate, a sheet of parafilm and a UV light source (Epileds, Tainan, Taiwan). Samples were irradiated at 365 ± 5 nm (200-220 mW/cm 2 ) 5 and 15 mm away from the surface of the light source, which provided dose of UV irradiation~17.2 J/cm 2 and~4.6 J/cm 2 , respectively. A sheet of parafilm was placed over the top of 96-well plate, and taped to the plate on all four sides. Each well was pressed to create a shallow groove. The plate was kept on ice before and during irradiation.
UV cross-linking of 4-thio-dU-elongated DNA primer (5'-33 P-labelled) to TdT was carried out after PEX. PEX were performed in Co 2+ -containing buffer as described above. Immediately after incubation, the reaction mixtures were chilled on ice and transferred as 10 µL drops to the wells on the parafilm tape. The ice container was placed underneath a 365 nm UV, so that the samples were 5 and 15 mm from the surface of the light source. Following incubation on ice the samples were irradiated at 365 nm for 5 min. Then the samples were transferred from parafilm wells to microtubes and quenched with 20 µL of loading solution. To verify the products of elongation reactions using 4-thio-dUTP and dUTP by TdT samples were examined on 15% polyacrylamide gel containing 8 M urea. Alternatively, to examine potential TdT-ON cross-links generated by irradiation, the samples were supplemented with SDS loading dye, heated for 5 min at 95 • C, and analyzed by electrophoresis on a 14% w/v SDS-PAGE gel. Proteins were stained with Coomasie Briliant Blue staining solution (Applichem, Darmstadt, Germany). TdT and radiolabelled-ONs cross-links were visualized using an FLA-5100 imaging system (FUJIFILM).

Conclusions
We have synthesized novel modified nucleotides that can be incorporated into DNA sequences by terminal deoxynucleotidyl transferase in the primer-extension reaction. We have succeeded in the enzymatic elongation by 4-thio-2'-deoxyuridine, N 4 -amino-2'-deoxycytidine and 2-pyridone base-bearing nucleotides utilizing template-independent TdT, which produced the artificial DNA fragments tailed with several to tens nucleotides at the 3 -end. For example, 4-thio-dU-label has potential in studying specific proteins which interact exclusively with the DNA 3'-end, hydrazine-bearing ONs could be successfully applied in a conjugation to improve in vivo delivery of therapeutic ONs or immobilization, whereas halogenated-ONs seem to be irreplaceable radiosensitizing agents. Enzymatic post-synthetic modification of ONs with various photo-and chemically-reactive groups is in progress towards further use for the modification of DNA or RNA. Thus, this study would open many possibilities for future applications, especially in the development of photo-labels, sensors, and self-assembling structures.