Patterns of the within-host evolution of human norovirus in immunocompromised individuals and implications for treatment

Abstract

Background: Currently, there is no licensed treatment for chronic norovirus infections, but the use of intra-duodenally-delivered immunoglobulins is promising; nevertheless, varying results have limited their wide use. Little is known about the relationship between norovirus genetic diversity and treatment efficacy.

Methods: We analyzed the norovirus within-host diversity and evolution in a cohort of 20 immunocompromised individuals using next-generation sequencing (NGS) and clone-based sequencing of the capsid (VP1) gene. Representative VP1s were expressed and their glycan receptor binding affinity and antigenicity were evaluated.

Findings: The P2 domain, within the VP1, accumulated up to 30-fold more non-synonymous mutations than other genomic regions. Intra-host virus populations in these patients tended to evolve into divergent lineages that were often antigenically distinct. Several of these viruses were widely resistant to binding-blocking antibodies in immunoglobulin preparations. Notably, for one patient, a single amino-acid substitution in the P2 domain resulted in an immune-escape phenotype, and it was likely the main contributor to treatment failure. Furthermore, we found evidence for transmission of late-stage viruses between two immunocompromised individuals.

Interpretation: The findings demonstrated that within-host noroviruses in chronic infections tend to evolve into antigenically distinct subpopulations. This antigenic evolution was likely caused by the remaining low immunity levels exerted by immunocompromised individuals, possibly undermining antiviral treatment. Our observations provide insights into norovirus (within-host) evolution and treatment.

Funding: Erasmus MC grant mRACE, the European Union's Horizon 2020 research and innovation program under grant agreement No. 874735 (VEO), and the NWO STEVIN award (Koopmans).

Keywords: Antigenic evolution; Chronic infections; Immunocompromised; Immunoglobulin; Norovirus; Virus evolution.

Fig. 1

Timeline of norovirus infections in immunocompromised patients. Each triangle represents a time point of sampling. Patients P17, P18 and P19 received antiviral treatments against norovirus. The quality of the NGS results was defined as “High” if at least 95% of the norovirus genome had a coverage ≥100×; “Medium” if at least 95% of the genome had a coverage ≥5×; and “Low” if neither of these criteria was met.

Fig. 2

Patterns of evolution of GII.4 (n = 10) and non-GII.4 (n = 5) noroviruses in immunocompromised individuals by Single nucleotide variant (SNV) analysis. (a) The number of the cumulative unique emerging SNVs (frequency ≥10%) per gene or protein domains by genotype (GII.4 and non-GII.4) and type of mutation (non-synonymous and synonymous). SNVs that were present at first day of collection (day 0) were not count for calculations. The median and the interquartile range are shown in red. P values were determined by comparing each gene to the correspondent RdRp. ∗P < 0.05, ∗∗P < 0.01 and ∗∗∗∗P < 0.0001. (b) Rate of mutations over time (cumulative unique emerging SNVs per site per day) in the genome of each individual patient by genotype and type of mutation. (c) Rate of mutations over time in the genome of aggregated patient data by genotype and type of mutation. The curves show LOESS fits, and shaded areas show 95% confidence intervals, as implemented in the geom_smooth function in ggplot2 R package. (d) Rate of mutations over time per gene of aggregated patient data by genotype and type of mutation. Only LOESS curves are shown per gene. (e) Analysis of the amino-acid (AA) position hotspots in GII.4 ORF2 and ORF3 genes. The x-axis indicates the AA positions of the protein relative to the GII.4 Sy 2012 reference (JX459908), while the y-axis indicates the frequency of patients (n = 10) in which the initial AA has changed at the consensus level (SNV frequency >50%). Positions with a frequency ≥0.3 (at least 3 patients) are shown as black circles. (f) GII.4 P-dimer structure (PDB: 5J35) showing both the blockade epitopes and the AA position hotspots of mutation. Blockade epitopes are colored based on panel e.

Fig. 3

Time-scaled phylogenetic tree for GII.4 noroviruses. The tree was constructed from clonal and NGS consensus VP1 nucleotide sequences of norovirus GII.4 from 14 chronically infected patients, along with reference sequences. Patient-derived sequences are indicated in the main tree. Green triangles indicate representative patient-derived VP1 sequences, while black triangles indicate reference VP1 sequences that were further expressed and characterized. Specific subtrees show the intra-host norovirus diversity within patients P1, P9, and P17. Posterior values above 0.7 are shown. Consensus sequences are shown as black circles. Matched highlighter plots show AA substitutions in the P-domain relative to the top sequence in the phylogeny. Trees for the non-GII.4 viruses are shown in Supplementary Fig. S5.

Fig. 4

Intra-host lineage dynamics based on the frequency of lineage-defining mutations over time. (a) Frequency of lineage-defining mutations to estimate lineage dynamics. Lineage-defining mutations of a (sub-)lineage were determined based on their prevalence in clonal sequences of such lineage. The frequency of each mutation was determined by SNV analysis, where only positions with a coverage ≥100×, Phred score ≥30, a frequency ≥1%, and at least three reads containing the SNV at the specific nucleotide position were considered. VP1 residue positions are given relative to the initial (day 0) VP1 consensus AA sequence derived from each patient. For all GII.4 sequences, the AA positions align with those of the Sydney 2012 reference sequence (JX459908). For patient P5, lineage-defining mutations of L2 are included within L3. (b) Norovirus P-dimer structure of the intra-host lineages. Residues shown in red and green denote within-host AA changes and indels between lineages of the same patient, respectively. HBGA glycans are shown in blue.

Fig. 5

Phenotypic characterization of patient-derived norovirus capsids. (a) Binding of GII.4-derived RLuc-VP1 fusion proteins to PGM-III and saliva containing HBGA. The luciferase activity of each RLuc-VP1 protein was adjusted to 5 × 10⁶ RLU/well before performing the binding assay. Plots are grouped by genotype and HBGA type (bottom table). The average RLU signal ± standard deviation of two independent experiments are shown. The presence of specific HBGAs in the 8 saliva samples (D1-D8) was tested by ELISA (Supplementary Fig. S3). Differences in binding of >4-fold (Δ) between intra-host-derived proteins are indicated. The HBGA-binding profile for non-GII.4-derived RLuc-VP1 proteins is shown in Supplementary Fig. S9. (b) Blocking activities of Ig preparations against different RLuc-VP1 antigens. Fourteen commercial Ig preparations were tested for their HBGA binding-blocking activity of patient and genotype control RLuc-VP1 proteins to PGM, using dilutions of the Ig preparations where the starting concentration was 2500 μg/mL. Plots are grouped by genotype, and each dot represents the titer of an Ig preparation versus an antigen. Ig preparations with non-blocking activity detected are shown as IC₅₀ = 5000 μg/mL. Only significant differences between intra-host-derived proteins are shown. ∗P < 0.05 and ∗∗∗∗P < 0.0001.

Fig. 6

Effects of antiviral treatments on norovirus intra-host populations. (a) Norovirus RNA levels in feces of immunocompromised individuals under antiviral treatment (n = 3). (b) Number of intra-sample SNVs (iSNVs: variants relative to the consensus sequence of the sample) over time. (c) Rate of de novo SNVs emerging over time. For panels b and c, only SNVs present at ≥10% of the reads were considered for the analysis.

Fig. 7

HBGA-binding and antigenic activities of P18-derived norovirus (GII.14) VP1 mutants. (a) P-dimer crystal structure of the P18_d0 consensus showing the localization of the three residues potentially involved in the resistance to Ig treatment. (b) Fold-change in the binding of RLuc-VP1 proteins to PGM-III and saliva samples containing HBGAs. Each graph shows the fold change in RLUs of the mutant proteins relative to their parental protein backbone. Plotted values are the average of two independent experiments. (c) Blocking activity of three Ig preparations (Ig-3, Ig-14, and Ig-15) against the RLuc-VP1 proteins.

Fig. 8

Phylogenetic evidence of person-to-person transmission between two immunocompromised individuals. (a) Maximum-Likelihood tree constructed from consensus (n = 34) and clonal (n = 201) sequences of the P-domain (768 nucleotides) derived from P19 and P20, along with GII.3 reference sequences (n = 60) from GenBank. (b) Time-scaled phylogenetic tree of the clonal sequences derived from P19 and P20. Selected posterior values >0.7 are shown.