Use of primer/template assemblies to examine transcription fidelity by T7 RNAP
Previous biochemical and structural data demonstrated that highly stable ECs may be formed by incubation of T7 RNAP with synthetic RNA:DNA (p/t) assemblies, and that the RNA primer is efficiently extended in the presence of the next (correct) incoming NTP (Temiakov et al., 2002
). In this work, we adapted this primer extension assay to examine transcription errors made by T7 RNAP in the presence of incorrect (non-cognate) NTPs, as has been done for DNAPs. The use of a common assay facilitates comparison of T7 RNAP with other members of the pol I family of nucleotide polymerases.
Stable T7 RNAP ECs were assembled on a nucleic acid scaffold in which a 32P-labelled 7 nt RNA primer was annealed to a template DNA strand having an unpaired downstream region. The ability of the RNAP to extend the primer following addition of substrate was then determined by gel-electrophoresis and autoradiography (). Consistent with previous results, T7 RNAP efficiently extended the RNA primer by 1 nt in the presence of the correct substrate (e.g., UTP for the p/t used in ; lane 2). Further extension of the 8 nt product due to misincorporation was also observed during the extended incubation time used in these assays (upper bands in lane 2). Incubation with the non-cognate nucleotides CTP and GTP resulted in much less efficient extension of the primer to 8 nt and beyond-(lanes 4 and 5) due to misincorporation events.
Extension of primer/templates in the presence of correct and incorrect nucleotides
Misalignment of the template strand as a potential mechanism for transcription errors
Unexpectedly, incubation with ATP (which is complementary to the downstream TSn+1 base on the p/t used in , and is therefore referred to as the +1 NTP) resulted in efficient extension by 2 nts, with little detectable extension by 1 nt (, lane 3). Two mechanisms that might account for this result are presented in .
In the misincorporation model, incorporation of AMP opposite the TSn base (A) would result in the formation of an A:A mismatch at the 3’ terminus of the RNA primer (, left). Efficient extension of the mismatch by correct incorporation of AMP opposite the next base (TSn+1, T) would result in rapid conversion of the 8 nt product to a final size of 9 nt, with little detectable accumulation of the intermediate product (i.e., the rate of mismatch extension is greater than that of misincorporation). While this mechanism can account for the failure to observe the intermediate product, it does not account for the preferential extension of the primer only in the presence of the +1 NTP vs other non-cognate NTPs (since misincorporation would have to occur in all cases as the initial, rate limiting step).
In an alternative model, it is proposed that T7 RNAP misaligns the template strand by rotating the TSn base out of the active site, allowing the formation of an A:T base pair at +1 (, right). Following incorporation of AMP, the template strand would be realigned and the mismatched 3’ terminus would be extended as above. This mechanism would account both for the preferential extension of the primer in the presence of the +1 NTP (due to the opportunity to base pair with the TSn+1 base) as well as the failure to detect an intermediate 8 nt product (due to rapid extension of the mismatched primer following realignment).
To differentiate between the misincorporation and misalignment models, we examined primer extension in different sequence contexts (, panels B-D). In each case the primer was more efficiently extended by 2 nts in the presence of the +1 NTP than by 1 nt in the presence of other non-cognate NTPs (compare lane 3 to lanes 4 and 5 for panels A-D) demonstrating the importance of an opportunity to form a base pair with the TSn+1 base. This was observed regardless of the nature of the mismatch that would occur opposite the TSn base (e.g., purine:purine, pyrimidine:pyrimidine, purine:pyrimidine mismatches).
Interestingly, for the p/t used in panel D, we observed extension by 1 nt in the presence of the correct substrate (UTP) followed by efficient extension for an additional 2 nts, to give a product of 10 nt (upper band in lane 2). On this p/t UTP is complementary to both the TSn and TSn+2 bases. Thus, after the primer is extended 1 nt by correct incorporation opposite TSn (lower band in lane 2 of panel D), incorporation of UMP opposite TSn+2 during the next two nucleotide addition cycles as a result of misalignment and subsequent mismatch extension results in further extension by 2 nts (upper band in lane 2 of panel D).
Effects of abasic sites in the template strand
To account for the preferential incorporation of the +1 NTP vs other non-cognate substrates, the misalignment model proposes the formation of a base pair between the +1 NTP and the TSn+1 base. We therefore asked whether introducing an abasic site at the TSn+1 position would suppress this effect (). As compared to a control template having C at TSn+1, which resulted in a high rate of primer extension by 2 nts in the presence of the GTP (lane 3), addition of the same substrate to a p/t template having an abasic site at TSn+1 resulted in inefficient extension, and by only 1 nt (lane 5). The abasic site at TSn+1 had little or no effect on the rate of incorporation of the correct substrate (UTP) or that of another incorrect substrate (CTP; ). Significantly, the rate of incorporation by GTP on the abasic template (leading to extension by 1 nt) was nearly identical to that of another non-cognate substrate (CTP) on this template or on the control template (), indicating that when the opportunity for base pairing at +1 is removed, incorporation of GMP is the result of misincorporation opposite the TSn base (as is the case for CMP). The use of a template having an abasic site at the TSn position resulted in a slight stimulation in incorporation of the +1 NTP (data not shown), suggesting that removal of the TSn base might facilitate misalignment by reducing steric constraints within the active site.
Evidence for a single NTP binding site
Recent observations with multisubunit RNAPs have suggested that binding of the +1 NTP in these enzymes might occur in a secondary site, without altering the register of the TSn base (Abbondanzieri et al., 2005
; Foster et al., 2001
; Gong et al., 2005
; Westover et al., 2004
), and that binding of this nucleotide might activate the complex, resulting in higher rates of correct incorporation opposite the TSn base (Foster et al., 2001
). Conceivably, binding of the +1 NTP and activation of the transcription complex by this mechanism might stimulate misincorporation
opposite the TSn base in T7 RNAP, which could also result in preferential extension of the primer in the presence of the +1 NTP.
To explore this, we compared the rates of primer extension due to incorporation of a different non-cognate nucleotide in the presence or absence of the +1 NTP (). As before, we observed a slow extension of the primer by 1 nt in the presence of the non-cognate substrate CTP alone (lanes 2-5) and rapid extension by 2 nts in the presence of the +1 NTP alone (GTP; lanes 6-9). In the presence of GTP and CTP together, the primer was rapidly extended by 2 nts to yield a product that is characteristic of extension by GG (the different mobilities of products extended by GG or CG in this gel system readily allow a discrimination as to which nucleotides are incorporated; compare lanes 13 and 14). Significantly, no extension by CG was observed under these conditions, indicating that rapid extension by 2 nts in the presence of both substrates is due to the sequential incorporation of two GMP residues, and not due to an enhanced misincorporation rate for CMP followed by efficient mismatch extension.
Evidence for a single NTP binding site
Efficient mismatch extension by T7 RNAP
In the experiments above there is little evidence for the formation of an intermediate product (extension by 1 nt) in the presence of the +1 NTP (, lanes 6-9) indicating that the efficiency of mismatch extension by T7 RNAP is relatively high, and is more rapid than misincorporation. To examine this, we compared the rate of extension of primers having a mismatch at the 3’ terminus to extension of primers having a correctly paired 3’ terminus (; UTP). In the same experiment, we also examined the rate of extension of a correctly paired primer due to misincorporation (in the presence of CTP) or due to incorporation of the +1 NTP (GTP). As predicted by the model, we found that the rate of extension of the mismatched primer in the presence of the correct NTP is very rapid (comparable to the extension of a correctly paired primer within the resolution of this experiment) and is much faster than extension due to misincorporation. Significantly, extension by the +1 NTP (which involves misalignment followed by mismatch extension) is also more rapid than misincorporation, again indicating that the pathway that leads to extension by 2 nts in the presence of the +1 NTP does not involve an initial misincorporation event (which would be rate limiting).
Mismatch extension and comparison of mechanisms of misalignment errors by T7 RNAP and pol I DNAPs
Comparison of error mechanisms of pol I DNAP vs T7 RNAP
The efficient extension of a mismatched primer by T7 RNAP is in contrast to results of previous studies with pol I DNAPs, which have shown that the presence of a mismatched base within the p/t reduces the rate of extension by factors of 100 to one million(Echols and Goodman, 1991
; Goodman et al., 1993
; Kunkel and Bebenek, 2000
). The slow rate of extension of a mismatched primer has important implications for fidelity and error correction by DNAPs, as it affects the balance between extension and editing due to exonuclease activity and other repair mechanisms (T7 RNAP is not known to have an editing/exonuclease function; (Huang et al., 2000
To facilitate comparison between the two systems, we examined primer extension by T7 RNAP and the Klenow fragment (KF) of DNAP in the same sequence context (). In contrast to the results obtained for T7 RNAP (right panel), no preferential incorporation of the +1 dNTP was observed for KF (compare left panel, lanes 7-11 for incorporation of the +1 dNMP (dGMP) versus lanes 12-16 for incorporation of another incorrect dNMP (dCMP). Moreover, the mismatched primer resulting from misincorporation of the +1 dNMP was poorly extended by KF, resulting in primer extension by only 1 nt (lanes 7-11). Thus, consistent with previous reports, the opportunity to form a base pair between the +1 dNTP and the TSn+1 base does not stimulate the misincorporation rate for pol I DNAP, nor is there a high rate of mismatch extension (Bebenek and Kunkel, 1990
; Kuchta et al., 1988
; Kunkel and Bebenek, 2000
; Wong et al., 1991
The differences for T7 RNAP and DNAP in their ability to extend a mismatch are likely to account for the different outcomes of the two systems as a result of a misincorporation event in the presence of the +1 NTP. Whereas for T7 RNAP the outcome is predominately a base substitution error (this work), previous studies indicate that for KF the predominant result is a deletion error (Bebenek and Kunkel, 1990
). The pathways that are thought to lead to these alternate outcomes are shown in . In the case of T7 RNAP (right panels) misalignment promotes the preferential incorporation of the +1 NTP as the initial step in the process. Subsequent realignment and efficient mismatch extension then “seals” this substitution into the nascent transcript. In the case of KF (left panels) the initial event is misincorporation of the +1 dNTP (apparently not
as a result of template strand misalignment, for there is no preferential incorporation of the +1 nucleotide; see , above). However, because a mismatched primer is poorly extended by DNAP, the situation is thought to be resolved by misalignment of the template strand, allowing correct pairing of the 3’ end of the primer with the next downstream base and subsequent extension, resulting in a deletion error (, left; (Bebenek and Kunkel, 1990
; Kunkel and Bebenek, 2000
; Kunkel, 2004
Effects of changes in the organization of the transcription bubble on frameshift fidelity
The p/t assemblies used in the experiments above are similar to those used to explore the fidelity of DNAPs (and were specifically chosen for this reason to allow comparisons between various systems). However, they do not have all of the features that are thought to be present in a transcription elongation complex. In particular, they lack a complementary NT DNA strand downstream from the active site. We therefore examined primer extension on p/t assemblies that included: i) no downstream NT strand (template 1); ii) a complementary downstream NT strand beginning at the n+2 position (template 2); or iii) a downstream NT strand beginning at the n+1 position (template 3) (). While the rate of extension in the presence of a non-cognate substrate (CTP) was the same on all templates (lower panel), the rate of extension by the +1 NTP (GTP) was slightly reduced on template 2, and dramatically reduced on template 3 (upper panel). The latter template requires melting of the duplex DNA at the n+1 position to allow base pairing with the +1 nucleotide (GTP). In this case, the rate of extension by the +1 NTP was comparable to misincorporation in the presence of CTP, indicating that the presence of a complementary NT base at the n+1 position suppresses misalignment. In separate experiments, we found that the presence of a complementary strand does not suppress extension of a mismatched primer (data not shown).
Effects of template topology on misalignment
We next asked whether ECs halted downstream from a promoter on a double stranded (ds) template would generate similar misalignment errors, and whether the frequency of these errors might be affected by lesions or mismatches in the NT strand (). Here, the sequence of each template was designed to allow the formation of an EC halted 15 nts downstream from the promoter (EC15) in the presence of GTP and ATP alone (due to lack of the next incoming nucleotide, UTP). Importantly, each template included a cytidine in the T strand at the TSn+1 position (+17), allowing detection of misalignment errors due to incorporation of GMP (GTP is present in the reaction used to generate EC15).
In the presence of a fully complementary NT strand (ds template) we observed inefficient extension beyond 15 nts, and only by 1 nt (lane 1), presumably due to a low level of misincorporation. However, in the absence of a complementary NT strand (templates pss-1 and pss-2) we observed efficient extension by 2 nts by the halted ECs (lanes 4 and 5). Strikingly (but consistent with the results above), efficient extension by 2 nts was also observed on templates having a mismatch or an abasic site in the NT strand at the +1 position (mismatch and abasic templates, lanes 2 and 3, respectively). As abasic sites and mismatches are the result of mutagenesis or DNA replication errors, these findings may have relevance to transcription fidelity in vivo under circumstances in which these alterations remain unrepaired prior to transcription.