Rapid insight into the host response in patients with COVID-19 or influenza at point of care with nanopore sequencing

Rebekah Penrice-Randal, a member of the Hiscox Lab (University of Liverpool, UK), began her presentation by noting that the she and her team are familiar with the nanopore sequencing of viral pathogens: they previously used the technology to sequence Ebola virus in the West Africa outbreak and Lassa fever virus in Nigeria. When the COVID-19 pandemic began, they set up a collaboration with ICECAP (University of Edinburgh, UK) to study the effect of the SARS-CoV-2 virus on tissues. Using post-mortem tissue samples, they were able to identify the presence of the virus in multiple tissues, but inflammation sites were mainly found in the lungs; from this, they were crucially able to identify that there was a virus-independent immune response in these tissues. Next, Rebekah and her team decided to research the effects of SARS-CoV-2 using blood collected at point-of-care from individuals diagnosed with SARS-CoV-2, in collaboration with the University of Southampton, UK. Finally, they have studied sequential SARS-CoV-2 and influenza infections in mouse models, in collaboration with the University of Liverpool, to study the potential impact of a seasonal influenza outbreak during the COVID-19 pandemic.

Rebekah and her colleagues are using a range of nanopore sequencing approaches to study SARS-CoV-2. Samples are sent to their Containment level 3 facilities at the University of Liverpool; RNA is then extracted and libraries prepared for sequencing. The group have utilised the ARTIC method of preparing and sequencing libraries of the SARS-CoV-2 genome, allowing analysis of its evolution over time. They are also performing nanopore metatranscriptomic sequencing of swab samples, to identify other species present and their possible impacts on disease in both fatal and non-fatal cases. Finally, they are using transcriptome sequencing to assess host response to SARS-CoV-2 or influenza in clinical research samples, and also in mice with sequential infections of influenza and SARS-CoV-2.

Rebekah outlined their nanopore sequencing pipeline when characterising host response in COVID-19 in human clinical research samples. Libraries are prepared using the cDNA-PCR Barcoding Kit (SQK-PCB109) – providing ‘a cost-effective way for a very quick insight’. She noted that samples are also sent off for short-read sequencing, but that this involves long waiting times due to the current queues, so they decided to determine whether they could generate the same information using nanopore sequencing. Samples were sequenced on the GridION or MinION Mk1C; in data analysis, the reads were aligned via minimap2, transcripts quantified using Salmon, and differential gene expression (DGE) performed with EdgeR in RStudio. She displayed principal component analysis plots for the transcriptomic datasets.

Comparing the clinical research samples taken at point of care showed that clusters for those with COVID-19 or influenza separated from that of the healthy controls. Looking at fatal vs non-fatal COVID-19 cases, these also clustered away from the healthy controls. Considerable overlap in transcriptional signatures was seen in the COVID-19 vs flu point-of-care sample clusters, and between the fatal vs non-fatal COVID-19 clusters; however, in each case, there were also some interesting differences between them. The team generated ~2.5 million reads per sample, from which they identified hundreds of upregulated and downregulated genes both when comparing COVID-19 or influenza vs healthy control samples. What they were most interested in, however, were the smaller sets of genes which were differentially expressed in COVID-19 vs influenza, and in fatal vs non-fatal COVID-19 cases. In COVID-19 vs influenza, 56 were upregulated and 1 downregulated; in fatal vs non-fatal COVID-19, 8 were upregulated and 72 downregulated. Interestingly, they found that the majority of differentially expressed genes between COVID-19 and influenza were immunoglobulin transcripts: 34 were impacted in both COVID-19 vs influenza and fatal vs non-fatal COVID-19 datasets. Rebekah highlighted one example in which levels of an immunoglobulin transcript in fatal COVID-19 cases was more similar to a healthy control than to the non-fatal COVID-19 group. They hypothesised that ‘an effective, adaptive immune response promotes survival in COVID-19.’

Rebekah then described the team’s work with mouse models. Here, they infected mice with influenza followed by SARS-CoV-2, to study the potential impact of a seasonal flu season on humans – luckily, a large flu season wasn’t seen in 2020. In this study, they sequenced more samples in multiplex, so generated ~1.5 million per sample, but found that this was sufficient to identify some differentially expressed genes. Here, they observed a sustained interferon and cytokine response in co-infected mice. Finally, they compared the human and mouse data to compare genes of interest. This revealed some overlapping genes to increase in abundance across the human lung, human blood, and mouse transcriptomes, whilst little overlap was seen in downregulated genes across the datasets.