Detection of RNA modifications and structure using nanopore sequencing - William Stephenson
William Stephenson of the New York Genome Center spoke about his research utilising direct nanopore sequencing to characterise base modifications in ribosomal RNA (rRNA).
Modified bases have been described in several major classes of RNA, including transfer RNA (tRNA), rRNA, and messenger RNA (mRNA), and these modifications provide structure and functionality beyond that found in the four canonical ribonucleosides.
Ribosomal RNA, in particular, exhibits extensive modifications in both prokaryotes and eukaryotes, and these modifications tend to occur in functionally important regions of the ribosome.
William first described how direct RNA nanopore sequencing was used to qualitatively and quantitatively characterise modifications in rRNA from E. coli and S. cerevisiae. The team utilised the standard nanopore direct RNA sequencing workflow, and, as rRNA is a significant fraction of RNA within cells, no enrichment steps were required. Heat maps of Q-score against read length revealed a large number of reads correlating to the known small and large ribosomal RNA subunit lengths for the two organisms being studied, indicating a high number of full-length reads.
The E.coli genome incorporates seven ribosomal RNA gene operons and William explained how characterisation of differential operon usage using short-read sequencing techniques is complicated by the high homology of these regions. Conversely, long-read direct nanopore RNA sequencing allows ‘straightforward assignment to the appropriate rDNA operon’.
The team used the Tombo analysis platform to align raw current signals from both native and in vitro transcribed rRNA to the reference rRNA sequence. In vitro transcription removes base modifications and provides a baseline sequence with which to compare the direct sequencing reads that include modified bases. As expected, the team observed deflections in current in the native sample at or proximal to known modification sites in ribosomal RNA. In addition, the team examined the dwell times (i.e. how long the molecule spends inside the pore), revealing that native 16S rRNA showed a large change in mean dwell time at position 1412, which is 10 nucleotides away from the known N4, 2′-O-dimethylcytidine (m4Cm) residue at position 1402. Using a Kolmogorov–Smirnov test (KS test) of current signal and dwell times of the native and in vitro transcribed samples, the team were able to comprehensively analyse the modification profiles across all nucleotide positions. These analyses were performed using both the Tombo and nanopolish tools with highly concordant results being observed.
It was shown that ribose methylation induces motor protein pause 10 nucleotides downstream of the methylated base, suggesting that modifications can measurably perturb motor protein kinetics. William suggested that this reflects the distance between the nanopore constriction/reader head and the motor protein.
Moving to the second part of his presentation, William discussed the utilisation of nanopore sequencing to detect RNA structure. SHAPE-Seq and the derivative technique SHAPE-MaP can be used to investigate RNA structure using short-read sequencing approaches; however, while useful, these approaches are limited by reverse transcription bias and the inability to interrogate long range interactions at the single molecule level. The technique relies on a chemical probe that covalently modifies the RNA in a structure-dependent fashion, stopping reverse transcription and allowing the RNA structure to be deduced through interrogation of the resultant cDNA sequences. Having unsuccessfully attempted SHAPE-seq in combination with direct RNA sequencing, the team hypothesised that the large adducts (on the 2’OH of the SHAPE probe) were causing issues with translocation through the nanopore.
After testing a variety of putative small adduct acetylation reagents, only one, 1-acetylimidazole (1AI), was found to display high reactivity to the ribose of flexible sites in RNA. Benchmarking this reagent in a standard short-read SHAPE-Seq assay revealed high correlation of results. Furthermore, the reagent was also found to be permeable to mammalian cell membranes. Applying the novel ‘nanoSHAPE’ workflow enabled the delivery of full-length reads. Interestingly, the 1AI modified RNA also exhibited increased dwell time 10 nucleotides downstream of the modification, confirming that the motor protein is responsible or pausing in the presence of ribose modifications. After further testing and optimization, the team intend to use this technique to investigate long-range structural interactions.
Summarising his presentation, William stated that ‘long-read information obtained by sequencing RNA directly enables the detection of multiple modifications at the single molecule level, providing information on phasing of modifications and expanding the ability to decipher full-length RNA structure’.