Template-independent DNA synthesis activity associated with the reverse transcriptase of the long terminal repeat retrotransposon Tf1


Amnon Hizi, Department of Cell and Developmental Biology, Sackler School of Medicine, Tel Aviv University, Tel Aviv 69978, Israel
Fax: +972 3640 7432
Tel: +972 3640 9974
E-mail: ahizy@post.tau.ac.il


Reverse transcriptases (RTs) possess a non-templated addition (NTA) activity while synthesizing DNA with blunt-ended DNA primer/templates. Interestingly, the RT of the long terminal repeat retrotransposon Tf1 has an NTA activity that is substantially higher than that of HIV-1 or murine leukemia virus RTs. By performing steady state kinetics, we found that the differences between the NTA activities of Tf1 and HIV-1 RTs can be explained by the substantially lower KM value for the incoming dNTP of Tf1 RT (while the differences between the apparent kcat values of these two RTs are relatively small). Furthermore, the KM values, calculated for both RTs with the same dNTP, are much lower for the template-dependent synthesis (TDS) than those of NTA. However, TDS of HIV-1 RT is higher than that of Tf1 RT. The overall relative order of the apparent kcat/KM values for dATP is: HIV-1 RT (TDS) > Tf1 RT (TDS) >> Tf1 RT (NTA) > HIV-1 RT (NTA). Under the employed conditions, Tf1 RT can add up to seven nucleotides to the blunt-ended substrate, while the other RTs add mostly a single nucleotide. The NTA activity of Tf1 RT is restricted to DNA primers. Furthermore, the NTA activity of Tf1 and HIV-1 RTs is suppressed by ATP, as it competes with the incoming dATP (although ATP is not incorporated by the NTA activity of the RTs). The unusually high NTA activity of Tf1 RT can explain why, after completing cDNA synthesis, the in vivo generated Tf1 cDNA has relatively long extra sequences beyond the highly conserved CA at its 3′-ends.


DNA-dependent DNA polymerase




long terminal repeat


murine leukemia virus


non-templated addition


primer binding site




reverse transcriptase


template-dependent synthesis


The basic structure and mechanism of propagation of the long terminal repeat (LTR) retrotransposons are similar to those of retroviruses. The enzyme reverse transcriptase (RT) of all retroelements synthesizes the double-stranded and integration-competent viral DNA by copying their single-stranded RNA genome. This complex reverse transcription process is catalyzed by the two enzymatic activities of the RT, the DNA-polymerase (that copies both RNA and DNA) and the RNase H, which concomitantly cleaves the template RNA strand in the DNA•RNA heteroduplex formed during reverse transcription [1]. DNA synthesis in retroviruses and most LTR retrotransposons is primed by a specific tRNA that is annealed at its 3′-end to a genomic RNA sequence, designated primer binding site (PBS) [2]. Subsequently, the complete double-stranded reverse transcription produced cDNA, which is flanked by two identical LTRs, is integrated into the host cell genomic DNA by the viral integrase (IN) [1,3]. In all studied retroviruses and LTR retrotransposons, the CA sequence at the 3′-ends of both cDNA strands is highly conserved.

Unlike most retroelements, where the RNA primers for initiating (–) strand DNA synthesis are specific cellular tRNAs, the LTR retrotransposon of Schizosaccharomyces pombe, Tf1, represents a unique group of eukaryotic LTR retroelements that use a distinct tRNA-independent self-priming mechanism. Here, the very 5′-end of the genomic RNA can fold back due to its complementarity with the genomic equivalent of the PBS; hence, there is formation of a duplexed RNA region. Consequently, when the terminal RNA segment (that in the case of Tf1 is 11 nucleotides long) is nicked, the 3′-end of the produced RNA segment, which is annealed to the PBS, can be extended by serving as an endogenous self-primer for DNA synthesis [4]. A similar self-priming mechanism for (–) strand DNA synthesis was also proposed for a variety of other related LTR retroelements, such as the highly homologous Tf2 as well as Maggy, Skippy, Cft-1, Boty, copia of maize and Tf1/shushi of vertebrates [5]. These retroelements, which belong to a single lineage of the Ty3/gypsy group of LTR retrotransposons [6,7], have probably diverged early in the evolution of LTR retrotransposons before retroviruses. Therefore, it is possible that the mechanism of self-priming represents an early form of initiating reverse transcription in retroelements [8]. Indeed, the recombinant Tf1 RT, expressed by us [9], was found to nick Tf1 genomic RNA at its 5′-end and produce in vitro an 11 nucleotide RNA segment that can then be extended by the DNA polymerase activity of the same enzyme upon the addition of dNTPs [10]. This finding substantiates the prediction that Tf1 RT is responsible for creating also in vivo a functional RNA self-primer. As could be expected, this functional self-primer is generated by the RNase H activity of Tf1 RT, since a mutant RT that lacks this activity cannot produce the self-primer [10].

Sequence analyses of terminal sequences of the in vivo generated Tf1 cDNAs have also suggested that there are distinct features of this retroelement’s reverse transcription, such as the inefficient removal of the RNA self-primer [11]. Based on these terminal sequences, substantial sequence additions, beyond the correct and conserved 3′-end CA dinucleotides, were identified [11,12]. This may imply that Tf1 RT has an exceptional capacity to add non-templated dNTPs to the cDNA ends. Indeed, we have found in a previous preliminary study that recombinant Tf1 RT is able to synthesize in vitro DNA, in a template-independent manner, by incorporating nucleotides at the 3′-end of the blunt-ended DNA strand [9]. In addition, the IN of Tf1 was found to efficiently remove relatively long sequences beyond the conserved CA 3′-end termini located at both strands of the cDNA [13]. Apparently, to efficiently integrate the Tf1 DNA into the cellular DNA, this Tf1 IN feature is required to compensate for the non-templated addition (NTA) (or terminal transferase) activity of Tf1 RT, since the added extra sequences may interfere with efficient integration.

Thus far, the relatively high NTA activity of Tf1 RT was shown in a preliminary study with a DNA oligonucleotides substrate that mimics the sequence of HIV-1 U5 LTR end [9]. In this assay system, Tf1 RT was more effective than HIV-1 RT in synthesizing DNA in a template-independent manner. In the study presented herein, we investigated the NTA activity of Tf1 RT in more detail, using DNA sequences derived from the authentic Tf1 genome. This was done in comparison with the NTA activity associated with two prototype retroviral RTs, the RT of the lentivirus HIV-1 and the RT of the gammaretrovirus murine leukemia virus (MLV). The results show substantial differences between the studied functions of the tested RTs. These differences may reflect some disparities between the RT-catalyzed intracellular reverse transcription processes.

Results and Discussion

The capacity of Tf1 RT, in comparison with the other studied RTs, to perform template-independent DNA synthesis was assayed with blunt-ended synthetic duplex DNA that mimics the U5 end of Tf1 LTR. In all tested cases, the 5′-end of the primer DNA strand, which has at its 3′-end the highly conserved CA sequence, was labeled with 32P and reaction products were resolved by high voltage urea/PAGE.

The NTA activity with a Tf1 LTR-derived substrate

We tested the non-templated extension of a blunt-ended DNA that mimics the U5 end of Tf1 cDNA (substrate set 37; see Materials and methods). Our previous study showed that the highest NTA activity with a different substrate (that mimics the U5 end of HIV-1 LTR) was obtained with Tf1 RT or HIV-1 RT using Mn2+ or Mg2+, respectively [9]. To test whether this rule applies also to the tested substrate set 37, and what the divalent cation preference of MLV RT is, we conducted the preliminary experiment shown in Fig. 1A. It is apparent also here that the NTA activity of Tf1 RT is substantially higher in the presence of Mn2+ compared to Mg2+, while HIV-1 RT prefers Mg2+. The NTA activity of MLV RT is quite weak with either of the divalent cations. However, since the majority of studies conclude that MLV RT prefers Mn2+ [1,2], we tested its NTA activity with this divalent cation in all the following experiments. It should be emphasized that the NTA activity of all RTs was tested with similar DNA-dependent DNA polymerase (DDDP) activities of the three RTs, as described previously [14] (see also Fig. S1). This gel shows the primer extension by these RTs with Mn2+ and Mg2+ for both Tf1 RT and MLV RT. Interestingly, unlike their NTA activities, the DDDP activities of these two RTs are quite similar with the two tested divalent cations. Figure 1B shows that the NTA activity of Tf1 RT prefers Mn2+ when tested also with a single dNTP as well as with a mixture of all four dNTPs.

Figure 1.

 The divalent cation preferences of the NTA activities of Tf1, HIV-1 and MLV RTs with the Tf1 U5 LTR-derived substrate DNA. (A) All NTA reactions were performed with similar RT associated DNA polymerase activities, calibrated as described in Materials and methods and shown in Fig. S1. In all experiments, the 32P 5′-end labeled primer in substrate set 37 was extended in the presence of either 0.5 mm MnCl2 or 5 mm MgCl2 with all four dNTPs present (each at a final concentration of 500 μm), and the resulting urea/PAGE analyses are shown. The asterisk indicates the 5′-end of the 32P-labeled primer. The 3′- and 5′-ends of the primer and template, respectively, are indicated by capital letters. (B) The NTA activity of Tf1 RT with either MnCl2 or MgCl2 (as in A) in the presence of either each dNTP or a mixture of all four dNTPs each at a final concentration of 500 μm. The gel shown in (B) was exposed for a longer time than the one shown in (A).

Figure 2 shows the products obtained with various dNTP combinations (each at a relatively high concentration). These experiments were performed with Tf1 RT (Fig. 2A) in comparison with the RTs of HIV-1 and MLV (Fig. 2B and C, respectively). As in Fig. 1, this assay was done with similar DNA polymerase activities of the three tested RTs, determined by template-dependent primer extension assays (see Fig. S1). It should be emphasized, however, that all NTA activities shown here strictly depend on the double-stranded substrate structures, as the single-stranded primers were never extended by any of the RTs tested (see Fig. 5C, later, for Tf1 RT, and data not shown). It is apparent from Fig. 2A that, when a single dNTP was present, there was a considerable primer elongation by up to six dNTPs in the case of dATP – see lane 3 (or even seven dATPs, detected after longer gel exposures; see lane 6 in Fig. 1B). Yet, most of the NTA products with dATP are 27 and 28 nucleotides long, indicating that the 24 nucleotides primer was extended by mainly three to four nucleotides (lane 3). In contrast, there is a weak single dNTP extension with dGTP and dTTP, while there is no detectable extension with dCTP (lanes 4–6). When two dNTPs are present in the reaction, all combinations containing dATP (lanes 7–9) result in primer elongation, while there is hardly any extension with dCTP+dTTP (lane 11). With dCTP+dGTP (lane 10) and with dGTP+dTTP (lane 12), there is a weak activity. Interestingly, the length of a small fraction of the products, generated with dATP+dTTP, is longer by two nucleotides than that produced with only dATP, adding together up to eight nucleotides (lane 9). Here, it is possible that these last two nucleotides may be Ts, that were added in a template-dependent manner, due to a potential backfolding of the nascent 3′-end tail (resulting from an A-T pairing). A careful comparison of the above results obtained with Tf1 RT with those shown previously for the same enzyme but with the substrate mimicking the HIV-1 U5 LTR end [9] indicates some differences. For example, the level of the NTA activity by all single dNTPs with the HIV-1 U5 LTR-derived substrate was very significant, whereas the NTA obtained with each dNTP (excluding dATP) with the Tf1-derived substrate is quite low (Fig. 2A, lanes 4–6). This implies that the sequence of the substrate DNA can affect to a certain extent the dNTP preferences of the NTA activity. We also noticed variations in the NTA levels when testing substrates with other terminal sequences, although the NTA activity of Tf1 RT was always higher than the comparative activity of both HIV-1 and MLV RTs (data not shown).

Figure 2.

 dNTP preferences for the NTA activities of Tf1, HIV-1 and MLV RTs with DNA derived from Tf1 U5 LTR. Reactions were conducted with the indicated P/T set and the selected dNTPs, each at a final concentration of 500 μm, as described in Materials and methods. The 3′- and 5′-ends of the primer and the template, respectively, are indicated by capital letters. The urea/PAGE analyses of all reactions are shown. (A) The NTA activity of Tf1 RT. (B) The NTA activity of HIV-1 RT. (C) The NTA activity of MLV RT.

For comparison purposes, the same Tf1-derived substrate was also used to test the NTA activity of HIV-1 RT and of MLV RT (Fig. 2B and C, for these two RTs, respectively). Evidently, the NTA activity of these two prototype retroviral RTs is lower than that of Tf1 RT. HIV-1 RT has a substantial NTA activity, although elongating the blunt-ended primer by mostly a single nucleotide (Fig. 2B). This result was obtained with dATP (lane 3), with dGTP (lane 5), with all dinucleotide combinations containing dATP (lanes 7–9), with dGTP+dCTP (lane 10) and with all three or four dNTP combinations (lanes 13–17). Similar to the results obtained with Tf1 RT, the strongest NTA activity of HIV-1 RT was obtained with dATP (lane 3), whereas, in contrast to Tf1 RT, there is also a substantial addition of dGTP that almost equals the NTA activity with dATP (lane 5). MLV RT has the least efficient NTA activity relative to the other two tested RTs (Fig. 2C), as it was able to extend the primer with only dATP (lane 3) or with dNTP combinations containing dATP (lanes 7–9). However, these extensions were relatively weak and, like most NTA products of HIV-1 RT, there was an elongation by just a single nucleotide.

In summary, with the specific substrate tested here, the order of an NTA with a single dNTP by Tf1 RT is dATP >> dGTP ∼ dTTP > dCTP. For HIV-1 RT the comparable order is dATP  ≥ dGTP  >  dCTP  ∼ dTTP (which is quite compatible with previous studies [9,15,16]), and with MLV RT this order is dATP >> dGTP ∼ dTTP ∼ dCTP. Generally, the extension by the purine dATP is markedly higher than by pyrimidines, and the only difference between these two RT groups is whether the NTA extension by the other purine, dGTP, is about as high as dATP (as with HIV-1 RT) or is very low or negligible, similar to the pyrimidines (as in the cases of Tf1 and MLV RTs).

Steady state kinetics of the NTA and template-dependent synthesis activities of Tf1 and HIV-1 RTs

To get insights into the steady state kinetics of dNTP incorporation by the NTA activity of Tf1 and HIV-1 RTs, we tested the initial template-independent primer elongation with the same substrate used above (set 37) as a function of increasing dNTP concentrations (Fig. 3). Since Tf1 RT has a significant NTA activity only with dATP, it was tested with this dNTP, while HIV-1 RT that can also incorporate dGTP was tested with either dATP or dGTP. These assays were performed with blunt-ended substrates that represent ‘standing start’ reactions (as opposed to the ‘running start’ reactions, described below). The data were quantified from several independent experiments, similar to the one shown in Fig. 3, and the double reciprocal linear plots of the initial velocity of primer elongation (expressed as fmol primers extended per minute) versus dNTP concentration were generated, with regression coefficients of 0.980–0.996 (not shown). All calculated steady state kinetics values are shown in Table 1.

Figure 3.

 Kinetics of dNTPs incorporation by the NTA activity of Tf1 and HIV-1 RTs. Reactions were conducted with P/T set 37 (Table 3), as described in Materials and methods. The RT amounts are given in the legend to Table 1. The reactions were initiated by the addition of the studied dNTP at various concentrations and incubated (at 37 °C) for 10 min. The reaction products were analyzed by urea/PAGE. (A) The NTA activity of Tf1 RT with increasing dATP concentrations. (B) The NTA activity of HIV-1 RT with increasing dATP concentrations. (C) The NTA activity of HIV-1 RT with increasing dGTP concentrations.

Table 1.   The steady state kinetics constants calculated for the NTA and TDS activities of Tf1 and HIV-1 RTs. The presented kinetic values were calculated from data similar to those shown in Figs 3 and 4. All activities were assayed with dATP, except for the last column performed with dGTP. The reactions were carried out for 10 min at 37 °C. Both RTs were calibrated by a primer extension assay to obtain similar DDDP activities; thus, we used 1350 fmol of p66/p51 heterodimeric HIV-1 RT and 2570 fmol of the p56 monomeric Tf1 RT (see Fig. S1). The calculated values of KM, or Vmax, were determined from the double reciprocal (Lineweaver–Burk) plots of dNTP concentration against the initial velocities of the primers extended. The plots were generated using linear regression analysis, all yielding regression coefficients around 0.98 (not shown). The kcat values are the steady state rate coefficients, calculated from the Vmax values, divided by the moles of each enzyme that was present in the reactions. The presented values were each calculated from three independent experiments.
KMm)9.12 ± 0.030.06 ± 0.02418.63 ± 170.4 0.007 ± 0.0018343.88 ± 116.6
Vmax (fmol·min−1)11.79 ± 0. 1210.64 ± 0.3411.1 ± 1.5520.75 ± 0.558.2 ± 0.91
kcat (min−1)8.7 × 10−37.8 × 10−34.3 × 10−3 8 × 10−33.19 × 10−3
kcat/KM (min−1·μm−1)0.95 × 10−30.130.01 × 10−3 1.150.0092 × 10−3

The calculated NTA-related values for Tf1 RT were the Michaelis–Menten constant (KM) of ∼ 9 μm dATP and the turnover number, kcat value, of 8.7  × 10−3 min−1. Consequently, the NTA efficiency value (kcat/KM) of Tf1 RT is 0.95 × 10−3 min−1·μm−1. Interestingly, the comparative values calculated for HIV-1 RT show that the KM value is dramatically higher by ∼ 46-fold, while the apparent kcat value is only about half of that of Tf1 RT. Therefore, the enzymatic efficiency of HIV-1 RT, kcat/KM, is greatly reduced relative to Tf1 RT by almost two orders of magnitude (Table 1). This result strongly suggests that the KM constant of the RT-DNA-dNTP complex plays a dominant role in determining the NTA activity, rather than the RTs’ turnover number (which is of the same order of magnitude as that of HIV-1 RT). This finding also explains why at very high dATP concentrations the overall level of the NTA activities of both RTs are quite similar (excluding the number of added nucleotides), while at low, close to physiological, concentrations Tf1 RT is by far more efficient than HIV-1 RT (Fig. 3).

To calculate the comparable steady state kinetic values for the template-dependent DDDP activity of Tf1 RT relative to HIV-1 RT, we performed the experiment shown in Fig. 4. Here, the template strand has four extra ‘T’s at its 5′-end to allow the extension, in a template-dependent manner, by four complementary ‘A’s at the 3′-end of the same primer used also in set 37 (substrate set 38). As expected, there is an efficient elongation of the labeled primer by four nucleotides as a function of dATP concentration, due to template-dependent synthesis (TDS). HIV-1 RT shows a full conversion of the primer to 28 nucleotide products, whereas Tf1 RT extends only a fraction of the primer with significant pausing. This indicates that, as shown earlier by us [9], the extent of the overall primer elongation by Tf1 RT is reduced relative to HIV-1 RT. As could be expected, at relatively high dATP concentrations there was a further extension of the primer beyond the four added ‘A’s, suggesting that NTA synthesis took place subsequent to the TDS. However, these additions are not due to ‘standing start’ reactions (shown in Figs 2 and 3) but rather to ‘running start’ reactions (as this NTA occurs after the RT reaches the end of the template, generated by TDS), namely the blunt-ended products of this initial TDS serve as the substrate for the subsequent NTA activity. The data, similar to those shown in Fig. 4, were quantified to calculate the steady state kinetics values for the TDS of the RTs (Table 1). However, it is difficult to derive the kinetic parameters for the second phase of the reaction (the ‘running start’ NTA), as it depends on the completion of the first TDS reaction. Hence, the steady state kinetics parameters for the ‘running start’ reaction were not calculated.

Figure 4.

 Kinetics of dATP incorporation by a TDS followed by the ‘running start’ NTA activity of Tf1 and HIV-1 RTs. The reactions were conducted with P/T set 38 (Table 3), as described in Materials and methods and in Table 1. The reactions were initiated by the addition of dATP in the indicated concentrations and incubated for 10 min at 37 °C. The TDS and NTA products are indicated. The reaction products were analyzed by urea/PAGE. (A) The NTA and TDS activities of Tf1 RT. (B) The NTA and TDS activities of HIV-1 RT.

For the TDS of Tf1 RT, the apparent values were KM = 0.06 μm dATP with a turnover number kcat of 7.8 × 10−3 min−1, and the apparent Tf1 RT efficiency kcat/KM is 0.13 min−1·μm−1. Taken together, it is conceivable that the NTA synthesis of Tf1 RT (Fig. 3A) is by far less effective than the TDS activity (Fig. 4A) due to the very large reduction in the combined kcat/KM value by ∼ 140-fold. This disparity results from a similar difference between the comparable KM values (with practically identical kcat values), suggesting that the KM value of Tf1 RT calculated for the same dNTP (in this case dATP) is markedly higher in the NTA reaction relative to TDS.

The comparison of the Tf1 RT calculated values with those of HIV-1 RT (Figs 3B and 4B) highlights, in addition to the differences in the NTA activity mentioned above, several other interesting and very significant differences (Table 1). First, the TDS efficiency, kcat/KM, of HIV-1 RT is about ninefold higher than that of Tf1 RT, indicating that the former is a substantially better DNA polymerase. This dissimilarity results from a similar quantitative difference in the comparable KM values, as the turnover numbers for the two RTs are practically identical. Second, the ratio between the comparable kcat/KM values of the TDS and the NTA (as mentioned above) is ∼ 140 for Tf1 RT while it is ∼ 115 000 for HIV-1 RT, indicating a staggering difference of ∼ 800-fold. Unlike Tf1 RT, the HIV-1 RT associated NTA activity can also utilize dGTP quite efficiently (Fig. 2B). Therefore, we have also evaluated the steady state kinetics of the ‘standing start’ reaction of HIV-1 RT with increasing dGTP concentrations (Fig. 3C). By and large, all kinetic values for dGTP are quite similar to those of dATP, namely the NTA-related KM values of HIV-1 RT for dATP and dGTP are almost indistinguishable (Table 1).

NTA activity with preformed 3′-end overhanging primers

The data shown in Figs 2–4 indicate that Tf1 RT can add several nucleotides to the blunt-ended primer/templates (P/Ts) via its NTA activity, whereas HIV-1 RT adds mostly a single nucleotide. One potential explanation is that Tf1 RT can elongate preformed 3′-end overhanging primers, while HIV-1 RT cannot extend such primers. To directly test this supposition, we used sets of P/Ts (all derived from set 37, used in Figs 2 and 3) with preformed one to five nucleotide additions to the 3′-ends of the primers (beyond the 5′-end of the bottom strand). To mimic the NTA sequences generated above with dATP, all overhangs were designed to contain only ‘A’s. However, there may be a potential pairing between these oligo ‘A’ overhangs and the three ‘T’s located at the 3′-end of the bottom strand (see set 37, Table 3). This may distort the analyses of the results due to potential ‘clamp’ formations ([14,17] and data not shown). To be more specific, we have presented recently in vitro evidence that RTs can perform template switches even with a very short (a minimum of two nucleotides) complementarity between the 3′-ends of the primer donor strand and the DNA or RNA template acceptor strands. These dinucleotide duplexes are markedly stabilized by the RT that ‘clamps’ together these otherwise unstable duplexes [14]. To prevent this possible clamp formation, the 5′-end of the primers (and consequently the 3′-end of the template) was shortened by three nucleotides (relative to the original set 37). Thus, a single nucleotide overhang was formed with a 22 nucleotides primer, a double nucleotide overhang with a 23 nucleotides primer and so on. The results, shown in Fig. 5A, indicate that Tf1 RT can elongate the primers in a template-independent manner even when the 3′-ends of the primers are unpaired. Thus, a primer with a single ‘A’ overhang was extended by up to three nucleotides (set 39). The two nucleotide overhanging primer was elongated by two nucleotides (set 40), and the primer with the three ‘A’s overhang was slightly elongated by a single nucleotide (set 41), while no extensions were evident with longer primer overhangs (sets 42 and 43). Apparently, there is a reverse correlation between the length of the primers’ overhangs and the efficiency of the NTA activity. Thus, with blunt-ended substrate (set 37), the activity is higher (with an addition of up to six to seven nucleotides) and, as the length of the primer’s tail increases, the NTA activity decreases. There is still a potential snag in the data interpretation, since in the processive NTA reaction with the blunt-ended substrate (set 37), six to seven consecutive ‘A’s were added, while with the preformed ‘A’ overhangs only three ‘A’s at most were added. One possible explanation could be that the template-independent addition is processive; therefore the already added ‘A’ can be elongated, as the polymerase active site of the RT is already situated upon the 3′-end of the primer. In contrast, the de novo binding of this RT site to preformed overhangs is less efficient, resulting in a less apparent elongation. It is also obvious from the data presented in Fig. 5C that none of the primers could be extended by Tf1 RT (as well as by HIV-1 RT, data not shown), when present as a single-stranded DNA (namely, with no bottom strand). This indicates that, in the case of all extended overhangs, the double-stranded parts of the P/Ts are required for proper NTA synthesis, although the 3′-end tails of these specific primers are unpaired.

Figure 5.

 The NTA activities of Tf1 and HIV-1 RTs with preformed primer overhangs. Reactions were conducted with the indicated P/T sets. All reactions were initiated with the addition of dATP, at a final concentration of 75 μm for Tf1 RT and at 500 μm for HIV-1 RT. (A) Tf1 RT with the various P/T sets. (B) HIV-1 RT with the various P/T sets. (C) Control reactions of Tf1 RT with the single-stranded primers that were included in the used sets.

As already shown above, unlike Tf1 RT, HIV-1 RT can efficiently add, by and large, mainly a single nucleotide to the blunt-ended substrate (see Figs 2B, 3B,C and 5B, set 37). The extensions of one nucleotide and two nucleotide overhanging primers (sets 39 and 40, respectively) by HIV-1 RT are very inefficient, while no significant activity is observed with any of the longer overhanging primers used (sets 41, 42 and 43) (Fig. 5B). This feature suggests that, like the TDS of HIV-1 RT [2,18,19], the NTA activity of HIV-1 RT is relatively distributive; thus the re-binding of the DNA polymerase active site of HIV-1 RT to the 3′-end of the nascent strand is very weak, even when this primer’s tail is overhung by a single nucleotide.

Effect of ATP on the NTA activity

The concentrations of ATP in cells are quite high and amount to about 3 mm, whereas the endogenous dNTP concentrations are far lower. It is known that ATP can participate in HIV-1 RT-mediated pyrophosphorolysis reactions of DNA with template overhangs, as well as in loosely binding the DNA polymerase active site of the RT [20–24]. Since ATP has been shown to inhibit the NTA activity of HIV-1 RT [25], it was of interest to test whether ATP can also affect the NTA activity of Tf1 RT. The results, shown in Fig. 6, lanes 1–7, imply that, like the NTA by HIV-1 RT, the NTA activity of Tf1 RT is sensitive to ATP at concentrations similar to the in vivo ATP concentrations. The two RTs did not show any NTA activity with ATP by itself, indicating that this activity is strictly a DNA polymerase. As suggested previously for HIV-1 RT [25], the most likely explanation for this phenomenon is that there is competition between ATP and dATP for the dNTP binding pocket of the enzyme, when ATP is in excess over dATP. Since the intracellular ATP concentrations are significantly higher than the dATP concentrations, it might be that a delicate interplay between these two nucleotides regulates the extent of intracellular NTA activity of the RTs. We are planning to study this interesting issue further.

Figure 6.

 The effects of ATP on the NTA and TDS activities of Tf1 and HIV-1 RTs. Tf1 RT (lanes 2–4 and 9–11) or HIV-1 RT (lanes 5–7 and 12–14) were assayed for their ‘standing start’ NTA activity (with set 37) (lanes 1–7), or for their TDS and the following ‘running start’ NTA activity (with set 38) (lanes 8–14). Lanes 1 and 8, P/T only (with set 37 or set 38 respectively). All other lanes are described in detail in the figure.

We have further investigated how ATP affects the TDS and the following ‘running start’ NTA activity of the two RTs (Fig. 6, lanes 8–14), obtained with set 38 (already used in Fig. 4). In the case of Tf1 RT, the TDS is only partially affected by ATP (lane 10), since this RT tends to pause more frequently, resulting in partially extended 25–27 nucleotide products (although the amount of full-length 28 nucleotide products is similar to that seen in lane 9). Nevertheless, the overall extension was even slightly higher, since more of the 24 nucleotide primer appears to be extended (as the 24 nucleotide substrate in lane 10 is reduced relative to lane 9). Still, the following NTA activity was inhibited by ATP. Interestingly, under the assay conditions employed at a high ATP concentration, ATP by itself can also be incorporated in a template-dependent manner (lane 11), suggesting that, like other RTs studied previously [23,26–30], Tf1 RT also has a low RNA polymerase activity. As to the effects of ATP on the TSD activity of HIV-1 RT, at first glance it seems that ATP does not inhibit the RT’s TDS, as the primer was fully extended in the presence of excess ATP over dATP (lane 13), similar to only dATP (lane 12). However, the control reaction with only ATP reveals that a full extension was also obtained (lane 14), indicating that, under these assay conditions, HIV-1 RT functions very well as an RNA polymerase. Moreover, a full TDS was obtained with lower ATP concentrations of even ∼ 0.4 mm ATP (data not shown). This implies that it is not clear whether the HIV-1 RT-directed DNA synthesis is inhibited by ATP. On the other hand, it is obvious that, under the same conditions, the DNA-dependent RNA polymerase activity of HIV-1 RT is substantially higher than the comparable activity of Tf1 RT.

One potential technical concern is that, at high concentrations, ATP might chelate the divalent cation used in the reactions, especially with Tf1 RT because of the lower Mn2+ concentration used. Consequently, such a reduction in the level of free divalent cation might perhaps indirectly lead to a reduced polymerase activity. Indeed, it was reported that there is some decrease (although not very strong) in HIV-1 RT Mg2+-dependent poly(rA)-oligo(dT)-directed synthesis (and a shift in the optimal Mg2+ curve) when ATP is present in the reaction [31]. In addition, others have found that ATP chelates Mg2+ better than Mn2+ [32]. To solve this possible difficulty in the Mn2+-dependent experiment, we tested the effects of increasing dATP concentrations on both the NTA and the TDS of Tf1 RT (assuming that dATP chelates divalent cations as well as ATP). It is apparent from the results, presented in Fig. S2, that under the employed assay conditions high dATP concentrations hardly affect the level of both NTA and TDS of Tf1 RT. Similar results were obtained with the comparable activities of HIV-1 RT (data not shown). This information removes potential technical doubts about the experiment shown in Fig. 6 and supports the conclusion that ATP has a direct effect on the RTs of Tf1 and HIV-1 at physiological concentrations.

Can RNA be extended by the NTA activity of Tf1 RT?

So far, all experiments done to assess the NTA activity of Tf1 RT were performed with DNA primers. Therefore, another question that is related to this study is whether Tf1 RT can extend RNA primers in a template-independent manner. This issue was answered with an RNA primer, annealed to its complementary DNA. To overcome the potential RNA hydrolysis by the RNase H activity of the RT, we used the D362N substitution variant of Tf1 RT that lacks the RNase H activity but retains its full DNA polymerase function [10]. This enzyme (designated RNase H-minus Tf1 RT) showed an NTA activity with a DNA primer (set 45, Table 3) which is similar to that of the wild-type Tf1 RT (Fig. 7B). However, the RNase H-minus Tf1 RT was devoid of any significant NTA activity with an RNA primer that was identical in its sequence to the tested DNA (set 44, Table 3) – see Fig. 7A. These results suggest that Tf1 RT can extend only blunt-ended duplexed nucleic acids containing DNA primers and is incapable of elongating the comparable duplexes with RNA primers.

Figure 7.

 The NTA activity of the wild-type and the RNase H-minus (D362N) substitution variant Tf1 RTs with RNA and DNA primers. The reactions were conducted with the indicated P/T sets, and the selected dNTPs, each at a final concentration of 500 μm. (A) Reactions conducted with the RNA-DNA substrate set 44 (see Table 3) and the RNase H-minus Tf1 RT: lane 1, P/T only; lane 2, P/T and RT; lanes 3–7, P/T and RT with the indicated dNTPs. (B) Reactions performed with the DNA-DNA substrate set 45 with either wild-type Tf1 RT (lanes 2–7) or with the RNase H-minus RT (lanes 8–13): lane 1, P/T only; lanes 2 and 8, P/T and RT; lanes 3–7 and 9–13, P/T and RT with the indicated dNTPs.

The data presented further confirm that one of the unique features of Tf1 RT is its high NTA activity relative to other RTs. A side-by-side comparison with HIV-1 RT implies that the main reason for this enhanced activity of Tf1 RT is that, under the NTA setting, Tf1 RT has a substantially lower steady state KM value for the incoming dNTP (Table 1). In contrast, the comparable value of HIV-1 RT in the template-dependent DNA polymerase reaction is much lower than that of Tf1 RT, indicating that HIV-1 RT is a better DNA polymerase than Tf1 RT. Previous in vivo data for Tf1 showed substantial sequence additions (beyond the correct and conserved 3′-end CA dinucleotides) [11,12]. Therefore, the in vivo relevance to the in vitro data on Tf1 RT presented herein is evident.

Several questions regarding the biological relevance of the high NTA activity of Tf1 RT may arise. The first is why does Tf1 RT need such an activity that is higher than that of other studied RTs? A possible answer is that the non-templated extra nucleotides, added by the RT to the 3′-ends of the fully synthesized cDNA molecules, are required to protect these ends from hydrolysis by non-specific cellular 3′-exonucleases. Alternatively, these sequences may be required to prevent potential aberrant annealing of the LTR 3′-ends (due to micro-homology to sequences in the opposite LTR) and the subsequent undesired DNA extensions. The obvious question is how other retroelements manage without such a high RT-related activity. As far as we know, the only RTs studied thoroughly for their terminal transferase activity (and shown to have a low NTA activity) are those of HIV-1 [15,16,25,33], of MLV in this study and of the retrotransposon Ty1 [34]. Therefore, it might be that other unstudied RTs still possess a high NTA activity. It might be as well that RTs from other retroelements do not need a significant terminal transferase activity, since the levels of the non-specific nucleases in their host cells are lower than those found in S. pombe. It may also be possible that other cellular factors contribute to the protection of the cDNA ends of these other LTR retroelements and hence there is no need to protect these ends by excessive additions of non-templated nucleotides.

It is well established that in order for the cDNA, produced by reverse transcription, to be integration-competent, the extra 3′-end cDNA sequence beyond the highly conserved 3′-end CA must be removed prior to integration. Indeed, after reverse transcription in infected cells is completed, two or three nucleotides are removed in many retroelements from the 3′-ends of the viral DNA by the 3′-end processing capacity of the INs [1,35]. We have found that, unlike other studied INs, Tf1 IN is capable of removing relatively long nucleotide extensions beyond the highly conserved CA sequence that is critical to the 3′-ends [13]. For that reason, it is likely that this high 3′-end processing activity of Tf1 IN can compensate for the high NTA activity of the RT by removing the RT-formed extra sequences. This special feature of Tf1 IN can also support the biological significance of the terminal transferase activity of Tf1 RT.

Taken together, the present study still leaves several open questions. For example, it is not clear why some dNTPs (mostly dATP) are incorporated by the RT’s NTA activity better than other dNTPs, or why several primer sequences are extended better than others by this activity. Moreover, we have shown in this study that the level of NTA activity among the tested RTs is Tf1 > HIV-1 > MLV. Therefore, another important question is what RTs’ sequences and structural elements shape the level of the NTA activity of a given RT.

Materials and methods


All DNA and RNA oligonucleotides used in this study were custom synthesized and purified by HPLC. Their sequences, lengths and designations are shown in Table 2. Likewise, the double-stranded substrates and their designations are shown in Table 3.

Table 2.   The synthetic oligonucleotides used in this study.
Oligonucleotide designationsSequence (5′–3′)
Table 3.   Primers/templates sets used in this study. The designations of the labeled primers are shown for each set, along with the compatible templates. The 5′- and 3′-ends of the primers and templates are indicated by capital letters. The overhangs at the 5′-end of the template strand (in set 38) are indicated by lower case letters. Asterisks indicate the 32P 5′-ends of the primers. Thumbnail image of

Reverse transcriptases

All recombinant RTs carry six histidine tags and were expressed and purified as previously described by us. These RTs were the p56 monomeric wild-type Tf1 RT [9] and the RNase H-minus Tf1 RT (the D362N substituted variant) [10], the wild-type BH-10 strain-derived p51/p66 heterodimeric HIV-1 RT [36] and the p70 monomeric wild-type Moloney MLV RT [37].

Template-independent and template-dependent DNA synthesis

The primers were 5′-end-labeled by T4 polynucleotide kinase and [γ32P]ATP, followed by heat inactivation and annealing to the appropriate templates (see Table 3), as previously described [38]. Reactions were performed with 26.4 nm of all P/Ts used and 11.2 nm of monomeric Tf1 RT, or with similar DDDP activities of the other RTs used, which were calibrated by primer extension assay, conducted as described previously [14,17]; see the presentation of similar activities in Fig. S1. The reactions were performed in 25 mm Tris/HCl pH 7.5, 43 mm NaCl, 4 mm dithiothreitol, 24 μg·mL bovine serum albumin, 4% glycerol and the divalent cations according to the RTs preferences, all in a final volume of 12.5 μL. Unless otherwise stated, Tf1 RT was assayed with 0.5 mm MnCl2, MLV RT with 0.8 mm MnCl2 and HIV-1 RT with 5 mm MgCl2. NTA reactions performed with the blunt-ended P/Ts were initiated by adding the dNTPs at the indicated combinations, each at a final concentration of 500 μm, followed by incubating for 30 min at 37 °C (unless otherwise stated). The reactions were then stopped by adding formamide loading buffer. The products were analyzed by 12% urea/PAGE followed by autoradiography, as previously described by us [37,38]. These reactions tested the ‘standing start’ NTA activity. TDS was assayed with P/Ts with 5′-end overhanging templates, followed by NTA that was initiated after the templates were fully copied (‘running start’ reactions). The incubations of all steady state kinetics experiments were performed for 10 min. The kinetics under these conditions were found to be within the initial reaction rates of catalysis (data not shown).


This research was supported in part by a grant from the Israeli Science Foundation (grant no. 411/07). A. Hizi is an incumbent of the Gregorio and Dora Shapira Chair for the Research of Malignancies.