Exploring a prolonged enterovirus C104 infection in a severely ill patient using nanopore sequencing

Abstract

Chronic enterovirus infections can cause significant morbidity, particularly in immunocompromised patients. This study describes a fatal case associated with a chronic untypeable enterovirus infection in an immunocompromised patient admitted to a Dutch university hospital over nine months. We aimed to identify the enterovirus genotype responsible for the infection and to determine potential evolutionary changes. Long-read sequencing was performed using viral targeted sequence capture on four respiratory and one faecal sample. Phylogenetic analysis was performed using a maximum likelihood method, along with a root-to-tip regression and time-scaled phylogenetic analysis to estimate evolutionary changes between sample dates. Intra-host variant detection, using a Fixed Ploidy algorithm, and selection pressure, using a Fixed Effect Likelihood and a Mixed Effects Model of Evolution, were also used to explore the patient samples. Near-complete genomes of enterovirus C104 (EV-C104) were recovered in all respiratory samples but not in the faecal sample. The recovered genomes clustered with a recently reported EV-C104 from Belgium in August 2018. Phylodynamic analysis including ten available EV-C104 genomes, along with the patient sequences, estimated the most recent common ancestor to occur in the middle of 2005 with an overall estimated evolution rate of 2.97 × 10^-3 substitutions per year. Although positive selection pressure was identified in the EV-C104 reference sequences, the genomes recovered from the patient samples alone showed an overall negative selection pressure in multiple codon sites along the genome. A chronic infection resulting in respiratory failure from a relatively rare enterovirus was observed in a transplant recipient. We observed an increase in single-nucleotide variations between sample dates from a rapidly declining patient, suggesting mutations are weakly deleterious and have not been purged during selection. This is further supported by the persistence of EV-C104 in the patient, despite the clearance of other viral infections. Next-generation sequencing with viral enrichment could be used to detect and characterise challenging samples when conventional workflows are insufficient.

Keywords: chronic infection; enterovirus; immunocompromised; intra-host evolution; nanopore sequencing; virus evolution.

Figure 1.

EV-C104 genome coverage. Coverage depth across EV-C104 genomes. The y-axis depicts the number of reads and the x-axis depicts the genome position. Red, single read in the reverse direction; green, single read in the forward direction. The open reading frame is depicted with blue arrows. Untranslated regions are depicted with red arrows.

Figure 2.

Phylogenetic reconstruction and genetic divergence of EV-C104. (A) Maximum likelihood phylogenetic tree inferred from EV-C104 complete/near-complete sequences. EV-C104 reference genomes were selected with ≥99 per cent query cover and >95 per cent percentage identity. Patient samples (in red with the letter E to indicate they have been enriched), ten complete EV-C104 genomes and three EV-C genomes from GenBank served as outgroups. CLC was used to generate the alignment and construct the tree. A General Time Reversible substitution model and a gamma distribution with invariant sites with 1,000 bootstraps were used. The emergence of two distinct genogroups A and B can be observed. Bootstrap values are shown at the branch nodes. (B) A maximum likelihood phylogenetic tree with a General Time Reversible substitution model and a gamma distribution with invariant sites with 1,000 bootstraps were generated without the outgroups. A root-to-tip regression with the y-axis corresponding to branch distances of the phylogenetic tree (in units of substitutions per site) and the x-axis corresponding to time (year). Patient sequences are in red (n = 4) and reference sequences (n = 10) are in blue. Points below the regression line indicate sequences that are less divergent than average (for their sampling date) and vice versa. Squares indicate sequences above the regression line and have more divergence; empty circles indicate sequences below the regression line and have less divergence. Abbreviations: Respiratory 1; R1, Respiratory 2; R2, Respiratory 3; R3, Respiratory 4; R4.

Figure 3.

Intra-host variants between patient samples. A Fixed Ploidy algorithm was used with 80 per cent variant probability and 75 per cent minimum frequency. To determine variation between sampling dates, Respiratory 2–4 were compared against Respiratory 1.

Figure 4.

Positive and negative selection pressure on individual codon sites. Only the patient genomes were used to analyse negative purifying selection (PP value 0.1). All EV-C104 genomes (n = 14) used in the regression analysis in Fig. 2b were used to analyse positive diversifying selection (PP value 0.1).