Project Summary
Human transcriptomes are endowed with many intriguing RNA modifications that are dynamic and reversible.
Many of these modifications regulate gene expression and cell-fate decision, while emerging new functions are
also likely to impact human health and disease. However, precise mapping and quantifying modifications
remains a challenge. While the Oxford Nanopore Technologies (ONT) platform of direct RNA sequencing can
detect RNA modifications as basecall errors, a major weakness is the lack of quantification of these errors and
the increased errors near the site of a modification. We hypothesize that using 3Dpol, the RNA-dependent RNA
polymerase (RdRp) of poliovirus, to copy each RNA provides a mechanism of “twice reading” of the RNA to
improve accuracy. In this twice-reading mechanism, 3Dpol captures the flexible fold-back of the 3'-end of an
RNA as the primer for replication, producing a double-stranded (ds)-hairpin helix that physically links the template
strand with the copied strand. This physical link is important, allowing each RNA modification to have two reads,
first through nanopore sequencing of the copied strand and then the template strand, such that the two reads
are relatable. Because the read in the copied portion is produced by the high-fidelity 3Dpol, it provides the
“ground truth” for the read in the template portion. In Aim 1, we will determine how 3Dpol copies an RNA
modification by defining its fidelity and signature of nucleotide incorporation. This is to determine how the two
reads are related to each other. Using a synthetic RNA template, we will measure the fidelity and the signature
of 3Dpol readout for abundant modifications in human mRNAs. We will perform the same analysis for processive
reverse transcriptases (RTs) to discern the differences between RNA- and cDNA-replication of an RNA. We will
also generate 3Dpol variants with a different nucleotide-incorporation signature than the native enzyme, which
will provide valuable new tools for quantifying RNA modifications. In Aim 2, we will determine the elongation
velocity of 3Dpol in end-to-end reading of long RNAs. This is to assess the processivity of 3Dpol upon
encountering challenging sequences and diverse structures typical of cellular RNAs. We will use a long non-
coding RNA to measure position-dependent velocity across the template. We will compare the velocity of 3Dpol
to processive RTs to provide new insight into the strengths and weaknesses of each enzyme. In Aim 3, we will
generate synthetic RNA standards, each with a site-specific modification (ψ, m6A, or m5C) for machine learning
and algorithm development. We will also generate a control library of long RNAs, each with a modification, and
determine the improved accuracy from 1-read to 2-read and even to multiple reads upon 3Dpol replication of the
library. We will then determine the improved accuracy upon 3Dpol replication of a human transcriptome. The
deliverable is a nanopore kit consisting of reagents for detecting an RNA modification in twice reading of
transcriptomes in a new strategy that will significantly improve read accuracy for research in all biological areas.