Development of an Agnostic Clinical Metagenomic Workflow for Respiratory RNA Viruses

Background Clinical metagenomics has the potential to provide an agnostic approach to respiratory virus diagnostics, but there are substantial methodological challenges. Viral loads vary by orders of magnitude among respiratory samples and nucleic acid extracts are dominated by non-target sequences, particularly for samples from the upper respiratory tract. This necessitates upstream virus enrichment and non-specific amplification approaches adaptable to high-throughput assays. We aimed to develop such a workflow for respiratory RNA viruses. Methods We used nasopharyngeal swabs in viral transport medium remaining after routine microbiological testing of clinical samples at Oxford University Hospital NHS Foundation Trust (UK) was completed. Pre-extraction virus enrichment and non-target nucleic acid depletion methods evaluated included differential centrifugation, bead-beating, nuclease treatment and virus precipitation. RNA was extracted, then reverse transcribed using random N ₉ priming to include the oligonucleotide (5’-GATGATAGTAGGGCTTCGTCAC-N ₉ -3’) used in subsequent amplification. Two amplification approaches were designed and compared side-by-side: isothermal multiple displacement amplification via Equiphi29 DNA polymerase using circularised template versus PCR-based sequence independent single primer amplification. Sequencing used the Oxford Nanopore Technologies platform. Results Our final optimised workflow combined multiple approaches to virus enrichment; sample clarification (3,200g centrifugation, 30 minutes), nuclease treatment, virus precipitation and centrifugation (13,200g, 30 minutes). RNA was extracted from the resultant pellet. Sequence independent single primer amplification was adopted in the final workflow as it consistently out-performed multiple displacement amplification. During development work, we detected a wide range of respiratory viruses, including coronaviruses (SARS-Cov-2, OC43, NC63, 229E), influenza A and B, respiratory syncytial virus, parainfluenza 2, 3 and 4, human metapneumovirus, human rhino- and enteroviruses, together with an extraction control (MS2) and controls for reverse transcription and amplification controls (Zika, murine respirovirus and orthoreovirus RNA). Conclusions We have developed an agnostic clinical metagenomics workflow for RNA respiratory viruses, with potential for use in molecular diagnostics. Subsequent work will determine sensitivity and specificity relative to performance current standard-of-care diagnostics such as commercial respiratory PCR panels.