Nanopore Dwell Time Analysis Permits Sequencing and Conformational Assignment of Pseudouridine in SARS-CoV-2

Abstract

Direct RNA sequencing for the epitranscriptomic modification pseudouridine (Ψ), an isomer of uridine (U), was conducted with a protein nanopore sensor using a helicase brake to slowly feed the RNA into the sensor. Synthetic RNAs with 100% Ψ or U in 20 different known human sequence contexts identified differences during sequencing in the base-calling, ionic current, and dwell time in the nanopore sensor; however, the signals were found to have a dependency on the context that would result in biases when sequencing unknown samples. A solution to the challenge was the identification that the passage of Ψ through the helicase brake produced a long-range dwell time impact with less context bias that was used for modification identification. The data analysis approach was employed to analyze publicly available direct RNA sequencing data for SARS-CoV-2 RNA taken from cell culture to locate five conserved Ψ sites in the genome. Two sites were found to be substrates for pseudouridine synthase 1 and 7 in an in vitro assay, providing validation of the analysis. Utilization of the helicase as an additional sensor in direct RNA nanopore sequencing provides greater confidence in calling RNA modifications.

Figure 1

Direct RNA sequencing for Ψ by monitoring current vs time traces in a protein nanopore-helicase platform. (A) Isomerization of U yields Ψ. (B) Structural depiction of the CsgG-helicase nanopore setup used in the R9.1.4 MinION/Flongle flow cells manufactured by ONT. (C) Example ion current vs time trace. This figure was made using the PDB files 4UV3(23) and 2P6R(24) that were selected on the basis of the description of this system in the literature and a patent.^,

Figure 2

Base-calling frequencies from direct RNA nanopore data when U or Ψ is present in biologically relevant sequence contexts. The bases are called for U or Ψ in (A) singly-, (B) doubly-, or (C) triply-modified contexts. The raw nanopore reads were base-called with Guppy v.4.0 followed by Minimap2 alignment, Samtools processing, and IGV visualization (n ∼ 24 000).⁻

Figure 3

Eligos2 analysis to locate Ψ in direct RNA sequencing data shows sequence context dependency in the magnitude of the oddR values. (A) Two example radar plots of ESB values for singly-modified Ψ versus U sites in RNA. (B) A plot of oddR values for the Ψ-modified sites in the RNAs studied. More example ESB plots and the reproducibility plots for the oddR values found for Ψ are provided in Figures S4 and S5.

Figure 4

Current intensity and dwell time analysis of RNA with U or Ψ passing through the protein nanopore sensor. (A) A representation of the helicase-nanopore sequencing setup to illustrate where the data are analyzed. Example (B) current histograms and (C) dwell time histograms for U or Ψ in the CsgG portion of the nanopore sensor, in which the distributions were analyzed with Nanocompore to identify sites of statistical differences between the populations by pairwise analysis using the Kolmogorov–Smirnov test. The P-values from the statistical test were −log transformed to visualize the results of the test by increasing the signal at those most different at each site based on (D) current and (E) dwell time. The analysis was conducted across 10-nt in which the modification could span the 5-nt window of the protein nanopore sensor region. The plots were constructed from >800 data points obtained from Nanopolish extraction of the currents and dwell times from the raw fast5 data files. More example histograms can be found in Figure S7.

Figure 5

Passage of Ψ through the helicase active site impacts the read dwell time compared to U that permits the detection of the epitranscriptomic modification at a distal site. (A) Schematic of the helicase-protein nanopore setup. Example (B) current and (C) dwell time histograms for a U or Ψ in the active site of the helicase. Stacked plots of −log(P-values) from the Nanocompore analysis for the complete passage of the suspect sites through the nanopore setup looking at statistical differences in the (D) current and (E) dwell times. (F) The median dwell times for the site most statistically significant based on Nanocompore analysis when it resides in the helicase. (G) Plot of the areas for the median dwell time distributions. The analyses were conducted on >1000 single-molecule measurements. Additional data are provided in Figure S9.

Figure 6

Analysis of nanopore sequencing reads for the SARS-CoV-2 RNA subgenomes for Ψ. (A) Interrogation of the SARS-CoV-2 RNA extracted from cell culture with modifications against an IVT-generated genome without modifications to find statistically significant differences in current intensity and base-calling error for the TRS-S subgenome. (B) As an example, U28 759 in TRS-3a yields a base-calling error found by Eligos2 and a long-range dwell signature found by Nanocompore/Nanopolish analysis. (C) Plot illustrating the Ψ sites found in the analysis. (D) RNAfold prediction of the region flanking U28 927 and U29 418 to illustrate the local secondary structure. Data for the other subgenomes is provided in Figures S11–20, and the full RNAfold analysis is found in Figure S21.