Deep splicing plasticity of the human adenovirus type 5 transcriptome drives virus evolution

Abstract

Viral genomes have high gene densities and complex transcription strategies rendering transcriptome analysis through short-read RNA-seq approaches problematic. Adenovirus transcription and splicing is especially complex. We used long-read direct RNA sequencing to study adenovirus transcription and splicing during infection. This revealed a previously unappreciated complexity of alternative splicing and potential for secondary initiating codon usage. Moreover, we find that most viral transcripts tend to shorten polyadenylation lengths as infection progresses. Development of an open reading frame centric bioinformatics analysis pipeline provided a deeper quantitative and qualitative understanding of adenovirus's genetic potential. Across the viral genome adenovirus makes multiple distinctly spliced transcripts that code for the same protein. Over 11,000 different splicing patterns were recorded across the viral genome, most occurring at low levels. This low-level use of alternative splicing patterns potentially enables the virus to maximise its coding potential over evolutionary timescales.

Characterisation of the transcriptome and proteome of SARS-CoV-2 using direct RNA sequencing and tandem mass spectrometry reveals evidence for a cell passage induced in-frame deletion in the spike glycoprotein that removes the furin-like cleavage site

“…Long read length sequencing using an Oxford Nanopore MinION on polyA+ RNA from cells infected with SARS-CoV-2 was used to characterise viral RNA species. This technique has recently been used to study herpes viruses, coronaviruses and adenovirus transcriptomes 14,26,27 . In total, 1,588,330 sequences were base called and passed QC, of those 527,401 were mapped to the BetaCoV/England/02/2020 genome using minimap2 ( Figure 1b).…”

“…A strict cut off of a 20 nt minimum polyA length was employed and 386,903 transcripts passed this test. Accurately determining the 5' end of direct RNAseq transcripts has been shown to be problematic 14,27,28 . The accepted model for coronavirus transcription (Figure 1a) proposes that all viral mRNAs have a common 5' end 29 and the analysis herein was further restricted to those transcripts that met this criterion which reduced the total number of identified full length subgenomic mRNA molecules to 72,124.…”

“…Extraction and sequencing was done as previously described 14 . Briefly, total RNA was extracted using TRIzol™ reagent (#15596026, Ambion) and the RNA extracted as per the manufacturer's recommendations but with 3 x 70% ethanol washes.…”

“…To cope with the very wide range of transcripts, and to enable grouping of transcripts into classes, our previously described ORF centric data analysis pipeline was utilised 14 . The transcripts were mapped to the viral genome with minimap2 and the mapping data was used to identify transcription regulatory sequences (TRS), the sites where the transcript meets the poly A tail at the end of the genome and junction sites which the software refers to as splice acceptor/donor sites as it was originally used to describe spliced adenovirus transcripts.…”

“…More recently we have focussed on the latest technologies to study the transcriptome of viruses, including direct RNA sequencing on nanopore devices using long read length sequencing. In particular, we developed a novel ORF-centric pipeline to analyse the very large amounts of transcriptomic data generated by this technique 14 . This pipeline was first used on data generated from adenovirus infected cells.…”