Off-season circulation and characterization of enterovirus D68 with respiratory and neurological presentation using whole-genome sequencing

Abstract

To explore an off-season enterovirus D68 (EV-D68) upsurge in the winter season of 2019/2020, we adapted a whole-genome sequencing approach for Nanopore Sequencing for 20 hospitalized patients with accompanying respiratory or neurological presentation. Applying phylodynamic and evolutionary analysis on Nextstrain and Datamonkey respectively, we report a highly diverse virus with an evolutionary rate of 3.05 × 10^-3 substitutions per year (entire EV-D68 genome) and a positive episodic/diversifying selection with persistent yet undetected circulation likely driving evolution. While the predominant B3 subclade was identified in 19 patients, one A2 subclade was identified in an infant presenting with meningitis. Exploring single nucleotide variations using CLC Genomics Server showed high levels of non-synonymous mutations, particularly in the surface proteins, possibly highlighting growing problems with routine Sanger sequencing for typing enteroviruses. Surveillance and molecular approaches to enhance current knowledge of infectious pathogens capable of pandemic potential are paramount to early warning in health care facilities.

Keywords: enterovirus D68; long-read sequencing; neurological infection; respiratory infection; whole-genome sequencing.

Figure 1

EV-D68 detection from January 2010 to March 2020 at the UMCG. (A) EV-D68 detection per month at the UMCG between 2010 and 2020. A distinct rise can be observed which does not follow the usual trend, shown in purple (2019) and red (2020). (B) Heatmap of EV-D68 detection per month at the UMCG between 2010 and 2020.

Figure 2

Time-scaled phylogenetic tree. The left panel indicates the corresponding countries to each EV-D68 sequence in the tree. The red arrows indicate the location of the four different clusters of our 2019/2020 patient samples. From the 892 references downloaded; 21 sequences were manually removed due to poor quality after the initial alignment (Supplementary Table S8). A total of 872 sequences were run through the Nextstrain pipeline. An additional 29 references were removed after a second alignment. The Nextstrain pipeline automatically removed problematic sequences during tree refinement (n = 12 references) and uploading to auspice (n = 7 references), this included EVD68_GR_05_02.12.19 (Supplementary Table S8). A total of 824 EV-D68 references were included in the phylogenetic analysis. Supplementary Figures S4, S5 illustrate zoomed in images of the clusters. The time-scaled tree had an estimated evolution rate of 3.05 × 10⁻³.

Figure 3

The frequency and distribution of single nucleotide variants per genome position between the patient samples (n = 20) and the Fermon strain (accession number AY426531). Every blue dot represents a non-synonymous change on the sequence, with the gene position depicted above. Sample frequency denotes the percentage of samples in the data set with the non-synonymous change. A Fixed Ploidy algorithm was applied on CLC Genomics Server v21.0.5 using the following parameters: 90% variant probability, 90% minimum frequency, min 100x coverage, max 100,000x coverage, Q-score threshold = 30.

Figure 4

Selection pressure on individual codon sites. Blue dots represent the selection pressure on individual codon sites for all reference genomes (n = 824), along with EV-D68 patients (n = 19; 1962–2020). Red dots represent the selection pressure on individual codon sites for only EV-D68 patients (n = 19). MEME was used to determine diversifying selection (PP value < 0.05). Individual sites are specifically indicated in Supplementary Figure S10.