Precision arbovirus serology with a pan-arbovirus peptidome

Abstract

Arthropod-borne viruses represent a crucial public health threat. Current arboviral serology assays are either labor intensive or incapable of distinguishing closely related viruses, and many zoonotic arboviruses that may transition to humans lack any serologic assays. In this study, we present a programmable phage display platform, ArboScan, that evaluates antibody binding to overlapping peptides that represent the proteomes of 691 human and zoonotic arboviruses. We confirm that ArboScan provides detailed antibody binding information from animal sera, human sera, and an arthropod blood meal. ArboScan identifies distinguishing features of antibody responses based on exposure history in a Colombian cohort of Zika patients. Finally, ArboScan details epitope level information that rapidly identifies candidate epitopes with potential protective significance. ArboScan thus represents a resource for characterizing human and animal arbovirus antibody responses at cohort scale.

Fig. 1. IgG profiling with the T7 phage displayed ArboScan library.

a ArboScan enables comprehensive quantification of antibody responses to arboviruses in humans and non-human animals. b Phylogeny of 98 flaviviruses in ArboScan which had full length genomes or full polyprotein sequences available. 15 of these viruses were included in VirScan. 11 additional flaviviruses without full genome sequences have proteins included in ArboScan. c Receiver operator curve evaluating VARscore as a predictor to distinguish viruses with moderate or high expected seroprevalence from viruses with low expected seroprevelence (viruses listed in Supplementary Table 3) in healthy United States residents (AUROC = 0.98). The ROC was generated using a sliding VARscore cut-off. d VARscores from serum anti-viral antibodies were compared to VARscores from anti-viral antibodies contained in processed mosquitoes harvested immediately post feeding, 24 h post feeding, and 48 h post feeding. e ArboScan was used to profile alpaca, goat, and horse antibodies before (Pre) and after arboviral exposure. VARscores for several alphaviruses, flaviviruses, and orbiviruses are shown. Boxes highlight the relevant virus with available PRNT data. Three alpacas and four goats were challenged with Mayaro virus (MAYV). Four horses were challenged with Japanese encephalitis virus (JEV). Three goats were challenged with Zika virus (ZIKV). a, d, and e were created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license. Source data are provided as a Source Data file.

Fig. 2. ArboScan profiles of Zika infected individuals.

a Acute and convalescent serum was collected from individuals who tested positive for Zika RNA. VARscores from flaviviruses were compared between individuals who had pre-existing flavivirus IgG (prior dengue) and those who did not (no prior dengue) at acute and convalescent time points (No prior dengue, Acute n = 8; Prior dengue, Acute n = 19; No prior dengue, Convalescent n = 8; Prior dengue, Convalescent n = 19). Boxes indicate median and interquartile range of each group. b Zika VARscore were compared to Zika plaque reduction neutralizing titer (PRNT) (Spearman correlation, rho = 0.61, p = 2.4 × 10^-6) and dengue VARscores were compared to corresponding dengue PRNT (Spearman correlation, rho = 0.49, p = 2.9 × 10^-8). c Prevalence of antibody reactivity to Zika epitopes was evaluated for each sample group. Zika infected individuals with or without pre-existing flavivirus antibodies generated antibodies to Zika non-structural protein (NS) 2a, NS2b, membrane glycoprotein precursor (Pr), capsid (C), and envelope (E). Source data are provided as a Source Data file.

Fig. 3. ArboScan profiles of individuals challenged with dengue.

a Longitudinal VARscores for dengue serotypes and representative flaviviruses for individuals challenged with single serotype dengue challenge strains (Dengue 1: day 0 n = 10, day 28 n = 10, day 90+ n = 10; Dengue 2: day 0 n = 11, day 28 n = 11, day 90 n = 10; Dengue 3: day 0 n = 10, day 28 n = 9, day 90 n = 9). Boxes indicate median and interquartile range of each group. b Dengue VARscores were compared to corresponding dengue plaque reduction neutralizing titer (PRNT). Dengue 3 VARscores correlate with dengue 3 neutralization (Spearman correlation, rho = 0.55, p = 0.006). Dengue 1 and dengue 2 VARscores do not correlate with dengue 1 and dengue 2 neutralization (Spearman correlation, rho = −0.17 and 0.39 respectively, p = 0.36 and 0.06 respectively) because (i) challenge with heterologous dengue serotypes produced increases in VARscores but not in neutralization and (ii) challenge with dengue 1 and dengue 2 induced variable immune responses. Source data are provided as a Source Data file.

Fig. 4. Correlations between dengue envelope epitope reactivities and dengue neutralization.

a Antibody reactivities to all dengue peptides were evaluated for correlation with dengue neutralization in the dengue challenge cohort. The vast majority of peptides that significantly correlated with dengue plaque reduction neutralizing titer (PRNT) were in envelope (E). Minimal epitopes that may explain E peptides were as follows: epitope 1: AAs 342-355, epitope 2: AAs 31-46, epitope 3: AAs 91-111, epitope 4: AAs 151−165, epitope 5: AAs 235-251, epitope 6: AAs 39-48, epitope 7: AAs 232-251, epitope 8: AAs 363-378, and epitope 9: AAs 394-407. b Antibody reactivity to dengue peptides that correlated with dengue neutralization in the Colombia Zika cohort. Epitopes 1, 3, and 5 were transferred from a. c Mature dengue envelope dimer with epitopes that correlate with neutralization highlighted in red. Source data are provided as a Source Data file.

Fig. 5. Breadth of fusion loop reactivity.

Antibody reactivity to fusion loop (FL) was evaluated for every flavivirus in ArboScan for all samples from the Colombia cohort. Only viruses with ≥3 reactive samples were included for visualization. Epitope icons were included for four sera of varying degrees of cross-reactivity. Jap. Enc. Japanese encephalitis, Mur. Val. Murray Valley, Duck Flavi. Duck Flavivirus, Mont. Myo. Leu. Montana Myotis Leukoencephalitis virus. b Confusion matrix of each individual by detectable Zika fusion loop antibodies and Zika neutralizing antibodies (two-sided Fisher’s exact test, p = 0.005). Source data are provided as a Source Data file.