Primer Cleavage Dependent PCR
Several coupled reaction schemes have been proposed for PCR in which a hybridization dependent primer activation step is linked to primer extension. In the pyrophosphorolysis-activated polymerization (PAP) assay [34, 35], a blocked 3'-terminal nucleotide is cleaved by attack of pyrophosphate (reverse of the polymerization reaction). For this to occur efficiently, high concentrations of pyrophosphate are required which may inhibit some polymerases. The range of blocking groups that can be accommodated at the 3'-terminus is very limited. A 3'-terminal dideoxynucleotide has been utilized in most studies [34, 35]. Of the four bases, only dideoxy-C can be readily incorporated using standard methods of oligonucleotide synthesis, limiting widespread use of this technique.
A coupled PCR assay has been proposed in which a blocked primer is cleaved after hybridization to the target sequence by a nicking restriction endonuclease . A restriction enzyme that has an asymmetric recognition sequence or that cuts only one strand at a hemimethylated site would be required to avoid cleavage of the template. To our knowledge, this reaction scheme has never been demonstrated experimentally. In any event, the requirement that the restriction enzyme recognition sequence be located near the 3'-end of the primer would severely limit the use of this method.
Use of both RNase H1 and RNase H2 to effect primer cleavage in a coupled PCR assay has been reported previously in the patent literature but minimally characterized [37, 38]. Unlike the Type II RNase H enzymes, Type I enzymes will not cleave a substrate having a single RNA residue. At least 3 consecutive RNA residues are required, and 4 for a high level of catalytic activity . Thus, use of a Type I RNase H in rhPCR would require that the primer have at least four consecutive RNA residues. This adds substantially to the cost and complexity of the synthesis of the primer and increases its susceptibility to degradation. The cleaved primer would terminate in two or more RNA residues which can inhibit primer extension and these RNA residues would be incorporated into the amplicon. Sagawa et al.  suggested that the specificity of Type II RNases H would be similar to that of a restriction enzyme and that cleavage, and hence amplification, would be completely prevented if there was a mismatch at the RNA:DNA base pair within the duplex formed between the primer and the template. Although this is not true, as seen in the present study, coupling RNase H2 cleavage to primer extension can be used to greatly boost the specificity of PCR.
Use of P.a. RNase H2 in rhPCR
The thermostability and temperature dependence of P.a. RNase H2 makes it well suited for use in rhPCR. The activity of the enzyme is unaffected by heating at 95°C for 45 minutes (Figure 2A). Critically, this level of resistance to heat inactivation is sufficient for use in PCR, even for reactions requiring an extended number of cycles and a sustained initial incubation at 95°C. To demonstrate this explicitly, P.a. RNase H2 was pre-cycled in PCR buffer for 80 cycles (without primers, dNTPs, or template), then all reaction components were added and rhPCR was performed (data not shown). The reaction efficiency was unchanged, indicating that the enzyme's thermal stability is sufficient to remain active throughout the range of use expected for all PCR applications. A further advantage of P.a. RNase H2 is that it is essentially inactive at temperatures below 30°C (Figure 2B and 2C). This effectively confers a hot start to reactions done using blocked-cleavable primers and any thermostable DNA polymerase. In addition, P.a. RNase H2 has sufficient activity at 50°C to support rhPCR, permitting the reaction to be performed throughout a broad temperature range (Table S2, Additional file 8).
Like other Type II RNase H enzymes, P.a. RNase H2 retains a high level of activity at concentrations of Mg2+ as low as 1 mM [[28–30]], enabling rhPCR to be performed at all magnesium concentrations typically employed in PCR. The enzyme is able to cleave heteroduplex substrates with a single ribonucleotide comprising any of the four bases. Cleavage also occurs at a riboA:deoxyU base pair (not shown), allowing the uracil-N-glycosylase (UNG) sterilization method to be used in rhPCR assays. Although reaction rates are slightly affected by the identity of the RNA base and the flanking DNA sequence, a single concentration of the enzyme and set of reaction conditions generally can be used regardless of the sequence of the target. The efficiency of primer cleavage is principally determined by the structure of the primer 3'-to the RNA base and the degree of complementarity present near the scissile linkage.
It is important to note that non-specific hydrolysis of the RNA linkage cannot lead to primer activation. Water catalyzed hydrolysis and enzymatic cleavage by contaminating single-strand specific ribonucleases both lead to the formation of a cyclic 2'-3'-phosphodiester at the 3'-terminus. This group blocks the 3'-end of the oligonucleotide and prevents primer extension. Spontaneous hydrolysis of the cyclic phosphodiester gives a mixture of 2'- and 3'-phosphate monoesters; enzymatic hydrolysis yields exclusively the 3'-phosphate. In either case, primer extension remains blocked. Background cleavage of the ribonucleotide linkage is problematic only if it occurs to such an extent that the amount of the primer is substantially depleted. Although divalent cations (including Mg2+) can facilitate water catalyzed hydrolysis of RNA containing oligonucleotides, especially at elevated temperatures , non-enzymatic degradation of primers containing a single RNA residue under thermocycling conditions used in PCR is negligible. If there is significant contamination of a sample with single-stranded ribonucleases, inhibitors such as human placental RNase inhibitor can be included in the reaction mixture as they do not affect the activity of RNase H enzymes. In our experience, this has not been necessary.
Recognition of Substrates having Base-Pair Mismatches by Type II RNase H Enzymes
RNase H2 plays an important role in the removal of RNA residues misincorporated into DNA due either to incomplete removal of RNA primers used to initiate DNA synthesis or polymerase errors [[41–45]]. Consistent with its role in DNA repair, Type II RNase H enzymes are also able to cleave substrates where there is an RNA:DNA base pair mismatch, but at a rate reduced compared to the corresponding perfect duplex [[31, 32, 46–48]]. For P.a. RNase H2, the rate of the reaction is decreased by about 10-fold (Figure 3). A decrease in rate of similar magnitude is seen with a mismatch on the 5'-side of the cleavage site (position "-1"). Mismatches at the "-3", "-2", and "+1" positions gave rise to smaller reductions in the cleavage rate. Outside of this region, effects of a base pair mismatch were negligible. In all cases, the only products observed by mass spectrometry, and by electrophoresis using radiolabeled substrates, reflected cleavage on the 5'-side of the RNA residue. More detailed kinetic studies of the effects of mismatches on cleavage rates are in progress.
Enhanced specificity of rhPCR
Coupling cleavage by RNase H2 to primer extension in rhPCR leads to greater specificity both with respect to template independent mispriming events (e.g., primer-dimer formation) and unwanted amplification of related sequences. The formation of primer-dimers is prevented even in assays that are very prone to this side reaction (Figure 5). This feature of rhPCR should be particularly beneficial in multiplex assays. The specificity of the assay with respect to misamplification of homologous sequences is far greater than can be achieved by PCR using unmodified primers. When there are mismatches over or neighboring the RNase H2 cleavage sites of both primers, the ΔCq values observed are extremely large. For the HRAS gene, the ΔCq between the rat and human sequences was greater than 50 cycles. This high degree of specificity should be very useful for the detection of low levels of heterologous DNA in xenogeneic transplant models (e.g., human tumors grown in a mouse host) and in other instances where there are related targets having closely spaced variations in sequence. In SNP detection, where it is necessary to exploit the effect of a single base pair mismatch on cleavage by RNase H2, the assay also shows far greater discrimination than can be achieved with standard allele-specific PCR.
Use of rhPCR for Genotyping
Several groups have employed RNase H in SNP discrimination assays using unbiased amplification of the target sequence linked to cleavage of an RNA-containing probe. Harvey, Han, and colleagues described the use of a thermostable RNase H1 enzyme in genotyping assays where cleavage of a fluorescence-quenched probe having four sequential RNA bases was used to discriminate base identity . The assay was coupled to PCR and cleavage of the probe occurred in real time during thermocycling, as in a 5'-nuclease assay. The Type II RNase H enzymes from Chlamydia pneumonia (C.p.) and from Thermus thermophilus (T.th.) have been used for SNP detection where a Molecular Beacon having a single RNA residue specific for the mutation site was cleaved following PCR in an end-point assay [[32, 48, 49]].
In rhPCR the discrimination between variant alleles relies on differential amplification of the matched and mismatched target sequences. The specificity of the assay is generally greatest when the RNA residue of the primer is placed over the SNP site. In a model system using "rDDDDx" blocked-cleavable primers, all 12 possible base pair mismatches were readily detected (Figure 7A). The average ΔCq was 10.9 using the same set of reaction conditions in each assay. With unmodified allele-specific primers, the average ΔCq was only 5.4.
The application of rhPCR to genotyping of genomic DNA samples was investigated with the SMAD7 rs4939827 (C/T) SNP locus. Individuals homozygous for the T/T allele are at increased risk for the development of colon cancer. With unmodified allele-specific primers, there was almost no discrimination between the two alleles (Table 2), precluding the use of traditional ASPCR for genotyping at this locus. With rhPCR, the two alleles were easily distinguished. Using "rDDDDx" primers with the RNA residue positioned over the SNP site, the ΔCq between matched and mismatched targets was approximately 12 for both the "T-allele" and "C-allele" specific primers. The three genotypes (C/C, C/T, and T/T) could be distinguished unambiguously. In a blinded study of 31 DNA samples representing different individuals, all were correctly identified. The assay was robust with both real-time and end-point modes of detection (Figure 8).
In studies with both a synthetic template (Figures S4 and S5, Additional files 6 and 7) and the SMAD7 SNP locus (Table 2), rhPCR primers with the mutation site located 5'-to the RNA base (position "-1") were less discriminatory than primers where the RNA base was placed over the SNP site. In the case of the SMAD7 locus, rhPCR primers with the SNP site at the "-1" position provided almost no differentiation between the two alleles, similar to unmodified allele-specific primers. At first, it might seem that placing the mutation site on the 5'-side of the RNA base would be optimal, providing discrimination both at the RNase H2 cleavage step and at the initiation of DNA synthesis. In contrast, with the RNA base opposite the mutation site (position "0"), the primer forms a perfect match to the template after RNase H2 cleavage, offering no opportunity for further distinction between the two alleles. However, in the former case, if primer cleavage and extension do occur on a mismatched template, the alternate allele is incorporated into the extension product and a new amplicon is created which is now a perfect match to the primer. As a result, a 10-fold decrease in the cleavage rate by RNase H2 can contribute at most a 3-4 cycle increase in the Cq value. On the other hand, if the RNA base is positioned opposite the mutation site, the mismatched template should be replicated faithfully. When this occurs, the effect of the mismatch becomes amplified with each cycle and produces a much greater increase in the value of ΔCq. For convenience, Additional file 9 contains a merged set of all of the additional files.
Other applications for thermostable RNase H2 in molecular biology
RNase H2 can also be used to increase the specificity of DNA ligation assays (data not shown). Oligonucleotide ligation assays (OLAs) employ two oligonucleotides (an acceptor oligonucleotide with a reactive 3'-hydroxyl group and a donor oligonucleotide with a 5'-phosphate) which are designed to hybridize adjacent to each other on a complementary target nucleic acid so that DNA ligase can join the two fragments [[50–54]]. The formation of this new longer species is detectable by a variety of means including PCR and fluorescent bead capture. An added degree of specificity for SNP detection relies upon the ability of DNA ligase to distinguish between a perfect match and mismatched base pair at or near the site of ligation. Donor and/or acceptor oligonucleotides can be designed that are modified to prevent ligation and contain an internal RNA residue near the ligation site. Blocking groups on the acceptor oligonucleotide useful to inhibit ligation are the same as those used to prevent primer extension. As with the blocked-cleavable primers used in rhPCR, cleavage of the blocked-cleavable OLA oligonucleotide will result in a free 3'-hydroxyl which can function as an acceptor for a 5'-phosphate during ligation. Cleavage of the donor oligonucleotide by RNase H2 will result in a 5'-ribophosphate, which will also function efficiently in ligation . This coupled reaction scheme could be employed to improve the specificity of OLAs for nucleic acid detection or genotyping.