Specificity Enhancement of PCR and Other DNA Polymerization Using Phosphorothioate-Modified dNTPs

  • Up to 100-fold Higher Specificity Demonstrated Using Alpha-Thio dNTP (dNTPαS) Analogs
  • Specificity Enhancement is Attributed to Attenuation of DNA Polymerase Kinetics
  • This Improvement Applies to PCR, Gene Cloning, and Nucleic Acid Detection, Including COVID-19 Viral RNA

DNA plays a central role in biology, and enzyme-mediated DNA polymerization of 2’-deoxynucleotide 5’-triphosphate (dNTP) monomers is one of the most thoroughly investigated processes in molecular biology. The applications of enzymatic DNA polymerization are numerous and powerful, extending to PCR, gene cloning, and nucleic acid detection. Collectively, these account for the steady increase in the huge volume of literature (>500,000 items) published over the past 50 years, as indicated by this chart.

Chart of the results for the search query ((("DNA polymerization") OR (PCR)) OR ("gene cloning")) OR ("nucleic acid detection") by Jerry Zon.

Despite the mechanistic scrutiny and the aforementioned successful applications, enzymatic DNA polymerization has been known to exhibit instances of misincorporation and mispriming that can lead to mutations and unwanted side-products. Multiple methods geared towards mitigating these problems have been published. Recently, Hu et al. reported remarkable enhancements in DNA polymerization specificity, which they achieved by simply including readily available phosphorothioate-modified dNTPs in the mixture of natural (aka cognate) dNTPs substrates. Before providing a synopsis of this novel methodology, some relevant introductory information about earlier studies of phosphorothioate-modified dNTPs in DNA polymerization will be discussed in the following section.


In 1979, Burgers and Eckstein reported that polymerization of 2’-deoxyadenosine 5’-O-(1-thiotriphosphate) (dATPαS) by DNA polymerase I from Escherichia coli (E. coli) incorporated the Sp but not Rp diastereomer, and that this diastereoselective process gave a polynucleotide with 3’-5’ internucleotide phosphorothioate linkages in the Rp configuration with overall inversion of configuration of phosphorus. 

Structure of 2’-deoxyadenosine 5’-O-(1-thiotriphosphate) (dATPαS) shown without absolute stereochemistry at the thiophosphate moiety and without negative charges on the thiotriphosphate moiety. Taken from TriLink BioTechnologies.

Sulfur-modified phosphorothioate linkage in DNA with Rp stereochemistry at phosphorus. Taken from commons.wikimedia.org and free to use.

Following the invention of PCR in 1983 by Kary Mullis (1993 Nobel Prize in Chemistry), dNTPαS analogs were shown to be useful in PCR-based applications such as DNA sequencing by either chemical degradation (Gish and Eckstein), exonuclease digestion (Labeit et al.), or pyrosequencing (Gharizadeh et al.). 

In the early literature for polymerase-mediated incorporation of dNTPαS substrates, the Zone found the following intriguing hint of improved performance relative to cognate dNTPs. Rienitz et al. reported that, in a phage-infected E. coli system, substituting dGTP with dGTPαS during in vitro replication led to an ~10-fold decrease in the frequency of G:G and G:T mispairs, i.e. ~10-fold improved fidelity, albeit not for the other three nucleotides. On the other hand, the Zone found a recent investigation by Pugliese et al. of single-molecule DNA polymerization by DNA Polymerase I (Klenow Fragment), wherein all four dNTPαS analogs were found to be incorporated 40 - 65% more slowly relative to the rate of corresponding native dNTPs. 

As will be evident in what follows, these seemingly unrelated effects of dNTPαS analogs on DNA polymerization are linked phenomenologically to the enhanced specificity of DNA polymerization using dNTPαS analogs discovered by Hu et al. and featured in this blog.

Specificity Enhancement of DND Polymerization for Sensitive Nucleic Acid Detection (Hu et al.)


In theory, enzymatic DNA polymerization should involve perfectly hybridized primer and template strands (i.e. no mismatched primer-template complexes) and perfect incorporation of cognate dNTP substrates (i.e. only A/T and G/C base pairs and no mispairing). In practice however, chemical reactions are not perfect and, in the real world, the existence of energetically accessible, alternative reaction pathways that lead to undesired side-products must be identified. Nature has dealt with this by enabling the evolution of in vivo enzymatic proofreading and misincorporation editing processes, but these are not simply mimicked in vitro.

Hu et al. cite multiple lead references to previous publications. In these, researchers faced major challenges for in vitro applications of enzymatic DNA polymerization due to primer-template mispriming and dNTP misincorporation, which can compromise the specificity of polymerization. This is especially true for PCR and other amplification processes, in which these events can accumulate during repeated rounds of polymerization. Similarly, primers can adventitiously serve as templates and thus generate so-called “primer dimers,” which can be much more complex than the simple case depicted here for PCR, especially in multiplexed systems involving many primers (e.g. Le et al.). 

Taken from commons.wikimedia.org and free to use.

Although primer design and various hot-start methods (e.g. CleanAmp®) can improve specificity, Hu et al. hypothesized that decreasing the rate of DNA polymerization might lead to specificity improvement. To investigate this possibility, and in view of prior work with diastereoselective enzymatic incorporation of Sp but not Rp dNTPαS, Hu et al. chose to study the effects of dNTPαS analogs as mixtures of Sp and Rp diastereomers, commercially available from TriLink.


Preliminary Comparison of DNA Polymerases:  As depicted here, initial studies involved 5’-FAM labeled DNA primer extensions on a short synthetic DNA template with four pairs of dNTPαS analogs, i.e. HPLC-separated Sp and Rp diastereomers, using either DNA polymerase I Klenow Fragment (Klenow with 3'−5' proofreading activity), Bst DNA polymerase Large Fragment (Bst), or Taq DNA polymerase (Taq) in the presence of the other three canonical dNTPs. Based on conventional denaturing PAGE analysis, full-length extension was obtained with the presumed Sp diastereomer, but not the Rp dNTPαS substrates, and Klenow and Bst were more efficient than Taq.

Drawn by Jerry Zon.

Reduction of the Incorporation Rate: To validate whether the dNTPαS analogs would exhibit reduced rates of incorporation, Hu et al. performed 5’-FAM labeled primer extension reactions with Bst for various times up to 180 seconds. The same primer was used with four templates with either T, dG, dC, or dA (shown here as N) for the first incorporating nucleotides, dATPαS, dCTPαS, dGTPαS, and TTPαS (also written as dTTPαS), respectively. PAGE analysis revealed that the incorporation of dNTPαS was generally slower than that obtained with the corresponding cognate dNTP, and dGTPαS was the slowest (~2-fold vs. dGTP), with the order of relative rates being TTPαS > dCTPαS > dATPαS > dGTPαS.

Drawn by Jerry Zon.

Suppression of Nonspecific DNA Polymerization: Having shown that dNTPαS can slow down DNA polymerization, Hu et al. next investigated the generation of primer dimers to determine whether specificity could be increased. As depicted here, they studied unlabeled DNA primer extension with Bst in the presence and absence of templates or/and primers. In the presence of only a template or primer, nonspecific DNA synthesis with dNTPs was observed by native PAGE, as indicated by the formation of multiple byproducts. In the presence of both a DNA template and primer, canonical dNTPs still caused nonspecific DNA polymerization and generated multiple byproducts along with the desired DNA product. 

Drawn by Jerry Zon.

Importantly, inclusion of either one, two, three, or all four dNTPαS analogs led to increasingly cleaner background in the presence or absence of a DNA template and/or primer. It was further noted that, “[i>

nterestingly, DNA polymerase [Bst>

was still active even in the presence of four dNTPαS triphosphates, while (sic) the yield remained the same.”

Suppression of Mismatched DNA Primer Extension: To test for dNTPαS inhibition of nonspecific extension under various base pairing conditions, Hu et al. designed a 5’-FAM primer and its matched, mismatched (5 mismatches W, X, etc.), and non-extendable templates (A−C, respectively), performing Bst extensions at various template−primer ratios and combinations. They found that the primer alone could generate nonspecific products, while the matched templates (A and C) suppressed nonspecific byproduct formation. Furthermore, the extendable template (A) afforded less nonspecific byproducts than the non-extendable template (C). Remarkably, when dCTPαS and TTPαS were used with canonical dNTPs, nonspecific extension was completely prevented. 

Drawn by Jerry Zon.

Next, to study dNTPαS discrimination between matched primer T/template A and mismatched primer T/template G systems, primer T extensions were performed in the presence and absence of dNTPαS. The results indicated that dCTPαS was incorporated on template A as expected, but not incorporated on template G, wherein the wobble T/G mismatch site was three nucleotides away. On the other hand, canonical dCTP was incorporated on template G, thus indicating higher discrimination of the dNTPαS.

Drawn by Jerry Zon.

G-T wobble. Taken from commons.wikimedia.org and free to use. Adapted by Jerry Zon.

This model construct for two-base-upstream T/G single-nucleotide discrimination was similarly applied by Hu et al. for real-time PCR (RT-PCR) using a plasmid containing the KRAS gene exon2 sequence and one of two different reverse primers (matched and T/G-mismatched) for two separate PCR reactions, each sharing the same forward primer (matched). In the presence of the matched reverse primer, the PCR Ct (threshold cycle) values using canonical dGTP (Ct = 6.8) and dGTPαS (Ct = 7.1) were almost identical. However, in the presence of the T/G-mismatched reverse primer, the Ct values with dGTP (Ct = 7.9) and dGTPαS (Ct = 12.3) were significantly different, indicating increased discrimination of the single-nucleotide mismatch by dNTPαS.

Suppression of Nonspecific PCR Amplification: The above findings led Hu et al. to explore dNTPαS in nucleic acid detection, DNA polymerization, PCR amplification, and gene cloning. They did so using complex templates, such as plasmid, human genomic DNA, and total human cDNA reverse-transcribed from cellular total RNA.

Although a higher template concentration caused nonspecificity, this adverse impact was successfully suppressed by the utilization of a dNTPαS. Briefly, seven pairs of primers for five gene targets and three types of templates (plasmid, human total cDNA, and genomic DNA) were examined with dCTPαS and Taq. dCTPαS could effectively inhibit nonspecific product formation, even for DNA amplification and cloning of full-length genes with thousands of base pairs from human total cDNA and genomic DNA.

Improving Accuracy and Sensitivity for COVID-19 Viral Detection: According to Hu et al., nucleic acid detection of SARS-CoV-2 RNA has been reported to have a high false-negative rate (10−70%), which presents a major problem for epidemic prevention and control. To address this analytical issue, Hu et al. took advantage of their aforementioned discoveries in high-specificity enhancement using dNTPαS substrates.

First, they performed RT-PCR analysis on SARS-CoV-2 RNAs based on SYBR-Green I in the presence of background RNAs (human total RNAs extracted from HEK293T cells). They also measured the high-resolution melting values and carried out agarose gel electrophoresis. Inclusion of dNTPαS in the reagents for original kits (supplied with cognate dNTPs) was shown to significantly increase specificity by eliminating byproducts formed with the original kits. Furthermore, and consistent with the suppression of nonspecific products and background noise, the original kits could detect down to ~250 RNA copies in the presence of human total RNA, whereas the dNTPαS-supplemented system enabled detection down to ~5 copies of viral RNAs, i.e. ~50-fold more sensitivity. Importantly, this improved performance was confirmed with actual clinical samples.


Although the above-mentioned results are factual data, the underlying mechanistic reasons for these remarkable effects involve speculation, at this time, that is summarized by Hu et al. as follows.

Since sulfur coordinates with Mg2+ weakly relative to oxygen, the Rp sulfur atom of the α-phosphate in Rp-dNTPαS coordinates poorly with the first and second Mg2+ ions (see Footnote) in the DNA polymerase active site, thus causing its non-recognition by the enzyme. In contrast, the Sp sulfur in Sp-dNTPαS perturbs only the third Mg2+ ion interaction (see Footnote), and its Rp-oxygen coordinates well with both the first and second Mg2+ ions, thus allowing Sp-dNTPαS recognition by DNA polymerase. Hu et al. state that this is consistent with their kinetic data for incorporation of Sp-dNTPαS, which was “just slightly slower than that of the canonicals.” 

The researchers further state that reduced rates of incorporation of primer extension using dNTPαS substrates relative to canonical dNTPs “can give a mismatched dNTP substrate sufficient time to dissociate from the extension site, thereby enhancing the specificity on DNA polymerization.”

Thus, they concluded that “the possible reasons for the specificity enhancement of DNA polymerization are (1) the reduced Sp−sulfur interaction with the third Mg2+, (2) the reduced electrophilicity at the P center [due to>

the S modification, (3) the dehydration by the S atom, and (4) the decreased polymerization rate.”

Closing Comments

The Zone was very impressed by Hu et al. and their novel findings, as they demonstrated remarkably “cleaner” DNA polymerization during amplification reactions using dNTPαS analogs in various model PCR-related applications, and especially RT-PCR of actual clinical samples involving COVID-19.

Two technical points of interest to the Zone that were not addressed in this publication are the extent of sulfur incorporation into the amplicons as Rp phosphorothioate linkages during PCR reactions and the fidelity of incorporation, whether it be a dNTP or dNTPαS substrate. The first point may affect template-polymerase interactions within the so-called “footprint” of the enzyme, while the second point is relevant to cloning efficiency, i.e. the relative amount of base mutations for dNTPαS-modulated vs. use of only cognate dNTPs. Hu et al. carried out Sanger sequencing of ensembles of PCR amplicons to measure fidelity, but the Zone believes that high-throughput sequencing of many individual (clonal) amplicons is needed to detect and quantify possible differences. 

What do you think?

As usual, your comments are welcomed. 


According to a review by Yang et al., contrary to the established theory that polymerase-substrate complexes and transition states have identical atomic composition and that catalysis occurs by the two-metal-ion mechanism, it has been discovered that an additional divalent cation has to be captured on the way to product formation. Unlike the canonical two metal ions (e.g. Mg2+), which are coordinated by DNA polymerases, this third metal ion is free of enzyme coordination. Its location between the α- and β-phosphates of dNTP suggests that the third metal ion may drive the phosphoryl transfer from the pyrophosphate leaving group opposite to the 3’-OH nucleophile. Experimental data indicate that binding of the third metal ion may be the rate-limiting step in DNA synthesis, and the free energy associated with the metal-ion binding can overcome the activation barrier to the DNA synthesis reaction.

This mechanistic view has been called into question by Perera et al., who used quantum mechanical/molecular mechanical calculations of polymerase. They found that a third magnesium ion positioned near the newly identified product metal site does not alter the activation barrier for the chemical reaction, indicating that it does not have a role in the forward reaction. This, they say, is consistent with time-lapse crystallographic structures following insertion of Sp dCTPαS, and although sulfur substitution deters product metal binding, this only has a minimal effect on the rate of the forward reaction.

Active site structure of the ternary complex of pol β reactant state from the final point of a lengthy well equilibrated molecular dynamics trajectory calculation. Important ångström distances are indicated (red dashed lines), as are metal coordination (black dashed lines). Two water molecules solvating the product metal make hydrogen bonds with phosphate oxygens of the primer terminal nucleotide and PΫ of the incoming dCTP (green dashed lines). Taken from Perera et al. and free to use. This work is written by (a) US Government employee(s) and is in the public domain in the US.

The mechanistic controversy continues, as discussed by Tsai in a commentary titled Catalytic mechanism of DNA polymerases—Two metal ions or three? The stated purpose of this article is not to examine the validity of prior studies but to address “what other experimental evidence is available, particularly from the viewpoint of enzymology, for such distinction.”

2 years ago
167 view(s)
Did you like this post?