Background:
Dissecting broadly neutralizing antibody (bnAb) specificity in polyclonal plasma in HIV-1 infection and post vaccination remains challenging. Delineation methods comparing plasma neutralization patterns against large virus panels with those of epitope-mapped bnAbs are widely used. Common to the current approaches is their inability to resolve the presence of multiple bnAb specificities, or, in rare occasions, of novel bnAb types. Here we report on virus panel classification, a novel delineation method designed to overcome these limitations and probed its utility on the XbnAb cohort, a large cohort of bnAb inducers (Nf304) identified in the Swiss 4.5K Screen (Rusert et al. 2016).
Methods:
Plasma neutralization fingerprints of bnAb inducers (Nf304) were evaluated using a 40-virus multiclade panel and compared to reference fingerprints of bnAbs (V1V2, V3 glycans, CD4bs, interface, fusion peptide, MPER) and broadly neutralizing DARPins (bnDs) (Friedrich et al. 2021). bnAb specificity was predicted using a maximum Spearman correlation method and the newly developed “virus panel classification” strategy. Creating thousands of random possible panels (size 10 to 35) from an existing data set, this method computationally identifies 20 to 30 virus sub-panels with optimal ability to classify bnAbs into their respective epitope clusters. Plasma specificity is predicted on each virus sub-panel using maximum Spearman correlation, plasmas clustered based on the percentages of prediction for each epitope in all virus panels and specificities defined based on the observed signatures in each plasma cluster.
Results:
Compared to the classical Spearman approach, we find agreement in 94% of plasmas with the panel method including only bnAbs as reference fingerprints, and 91% of plasmas when additionally integrating bnDs. The virus panel classification allowed a more detailed allocation of bnAb specificity and identified bnAb sub-groups with V3 activity. Most notably, we identified 77 (bnAb reference) and 85 (bnAb plus bnD reference) patients with potentially multi-specific bnAb activity exhibiting MPER features in combination with V1V2, V3 glycan, and/or CD4bs patterns.
Conclusions:
Application of the new viral panel classification to the XbnAb cohort provided a highly reliable prediction of bnAb specificity. In addition, the method provides insight into potential multi-specific bnAb responses, indicating its usefulness for analyzing bnAb responses in future vaccine trials.