Chemical modifications to DNA, such as 2' modifications, are expected to increase the biotechnological utility of DNA; however, these modified forms of DNA are limited by their inability to be effectively synthesized by DNA polymerase enzymes. Previous efforts have identified mutant Thermus aquaticus DNA polymerase I (Taq) enzymes capable of recognizing 2'-modified DNA nucleotides. While these mutant enzymes recognize these modified nucleotides, they are not capable of synthesizing full length modified DNA; thus, further engineering is required for these enzymes. Here, we describe comparative biochemical studies that identify useful, but previously uncharacterized, properties of these enzymes; one enzyme, SFM19, is able to recognize a range of 2'-modified nucleotides much wider than that previously examined, including fluoro, azido, and amino modifications. To understand the molecular origins of these differences, we also identify specific amino acids and combinations of amino acids that contribute most to the previously evolved unnatural activity. Our data suggest that a negatively charged amino acid at 614 and mutation of the steric gate residue, E615, to glycine make up the optimal combination for modified oligonucleotide synthesis. These studies yield an improved understanding of the mutational origins of 2'-modified substrate recognition as well as identify SFM19 as the best candidate for further engineering, whether via rational design or directed evolution.

DNA is a foundational tool in biotechnology and synthetic biology but is limited by sensitivity to DNA-modifying enzymes. Recently, researchers have identified DNA polymerases that can enzymatically synthesize long oligonucleotides of modified DNA (M-DNA) that are resistant to DNA-modifying enzymes. Most applications require M-DNA to be reverse transcribed, typically using a RNA reverse transcriptase, back into natural DNA for sequence analysis or further manipulation. Here, we tested commercially available DNA-dependent DNA polymerases for their ability to reverse transcribe and amplify M-DNA in a one-pot reaction. Three of the six polymerases chosen (Phusion, Q5, and Deep Vent) could reverse transcribe and amplify synthetic 2′F M-DNA in a single reaction with <5 × 10–3 error per base pair. We further used Q5 DNA polymerase to reverse transcribe and amplify M-DNA synthesized by two candidate M-DNA polymerases (SFP1 and SFM4–6), allowing for quantification of the frequency, types, and locations of errors made during M-DNA synthesis. From these studies, we identify SFP1 as one of the most accurate M-DNA polymerases identified to date. Collectively, these studies establish a simple, robust method for the conversion of 2′F M-DNA to DNA in <1 h using commercially available materials, significantly improving the ease of use of M-DNA.

Chemical modifications can enhance the properties of DNA by imparting nuclease resistance and generating more-diverse physical structures. However, native DNA polymerases generally cannot synthesize significant lengths of DNA with modified nucleotide triphosphates. Previous efforts have identified a mutant of DNA polymerase I from Thermus aquaticus DNA (SFM19) as capable of synthesizing a range of short, 2'-modified DNAs; however, it is limited in the length of the products it can synthesize. Here, we rationally designed and characterized ten mutants of SFM19. From this, we identified enzymes with substantially improved activity for the synthesis of 2'F-, 2'OH-, 2'OMe-, and 3'OMe-modified DNA as well as for reverse transcription of 2'OMe DNA. We also evaluated mutant DNA polymerases previously only tested for synthesis for 2'OMe DNA and showed that they are capable of an expanded range of modified DNA synthesis. This work significantly expands the known combinations of modified DNA and Taq DNA polymerase mutants.

Modified‐DNA polymerases have been evolved that can synthesize long strands of modified oligonucleotides. The resulting modified DNA (M‐DNA) has many high‐value potential applications, such as clinical diagnostics and therapeutics. To fully apply M‐DNA, it must be reverse transcribed back into natural DNA. Previously, error prone reverse transcriptases have been required for this step, and then the resulting DNA has to be amplified in a separate amplification reaction. Using a synthetic template with 2′F modified nucleotides, we tested a panel of commercially available DNA‐dependent DNA polymerases for their ability to reverse transcribe (RT) and amplify M‐DNA in a one‐pot reaction. We found that four of the six polymerases chosen (Phusion, Q5, Deep Vent, and Vent) were able to RT and amplify synthetic 2′F modified DNA in a single reaction. These products were sent in for high‐throughput sequencing and it was determined that Q5, Deep Vent and Phusion had between 90 and 95% matched reads, while Vent had less than 85%. The conditions of this process (reaction buffer and PCR cycling conditions) were also shown to affect the accuracy, but to a lesser degree. These polymerases were also able to RT and amplify enzymatically synthesized 2′F templates, allowing us to use this as a tool to quantitatively determine error rates of M‐DNA polymerases using high‐throughput sequencing. This demonstrated that P1, one of our labs best modified‐DNA polymerases, was almost four times more accurate than SFM4‐6, the fields previous best enzyme. These discoveries will enable the wider use of 2′F M‐DNA, as well as add to the. evidence supporting the use of commercially available DNA polymerases for reverse transcription reactions. Support or Funding Information National Science Foundation

Recently, several modified‐DNA (M‐DNA) polymerases have been identified which can synthesize long M‐DNAs. While these discoveries have enabled new applications of M‐DNA, these enzymes typically have poor fidelity, possessing error rates several orders of magnitude worse than native enzymes. However, to date, efforts to both quantify error rate, as well as experimental approaches towards understanding the origin of the poor fidelity of M‐DNA polymerases, have been limited. Here, we use a high‐throughput sequencing assay to characterize the error rate of leading M‐DNA polymerases, which showed that our enzyme, P1, has an error rate 1.6 times lower than SFM4‐6, the most frequently used Taq mutant for 2′M‐DNA synthesis. We also characterize the importance of individual mutations on fidelity, beginning to experimentally address the origins of fidelity, and have found that reverting even one mutation in a mutant yielded a fidelity that was 2.4 times greater than its parent, which will help us improve future M‐DNA polymerase rational design. Support or Funding Information National Science Foundation