Signatures of selection in the human antibody repertoire: Selective sweeps, competing subclones, and neutral drift

Abstract

Antibodies are created and refined by somatic evolution in B cell populations, which endows the human immune system with the ability to recognize and eliminate diverse pathogens. However, the evolutionary processes that sculpt antibody repertoires remain poorly understood. Here, using an unbiased repertoire-scale approach, we show that the population genetic signatures of evolution are evident in human B cell lineages and reveal how antibodies evolve somatically. We measured the dynamics and genetic diversity of B cell responses in five adults longitudinally before and after influenza vaccination using high-throughput antibody repertoire sequencing. We identified vaccine-responsive B cell lineages that carry signatures of selective sweeps driven by positive selection, and discovered that they often display evidence for selective sweeps favoring multiple subclones. We also found persistent B cell lineages that exhibit stable population dynamics and carry signatures of neutral drift. By exploiting the relationship between B cell fitness and antibody binding affinity, we demonstrate the potential for using phylogenetic approaches to identify antibodies with high binding affinity. This quantitative characterization reveals that antibody repertoires are shaped by an unexpectedly broad spectrum of evolutionary processes and shows how signatures of evolutionary history can be harnessed for antibody discovery and engineering.

Trial registration: ClinicalTrials.gov NCT02987374.

Keywords: Adaptive immunity; Population genetics; Somatic evolution.

Conflict of interest statement

The authors declare no conflict of interest.

Copyright © 2019 the Author(s). Published by PNAS.

Figures

Fig. 1.

Dynamics and molecular features of antibody repertoires. (A) Schematic of experiment design. (B) Dynamics of antibody repertoires. Each line represents a clonal B cell lineage, and its width indicates the fractional abundance of that lineage (the number of unique sequences belonging to the lineage divided by the number of unique sequences in the entire repertoire) at a given time. Colors indicate distinct lineages (colors are reused across panels corresponding to different subjects and do not indicate shared sequences across subjects). The most abundant 500 lineages within each subject’s repertoire at D7 are shown. (C) Dynamics of vaccine-responsive lineages. (D) Dynamics of persistent lineages. (C and D) Each line represents a clonal lineage. (E) Distributions of somatic mutation density within the V gene in sequences belonging to vaccine-responsive lineages, persistent lineages, or the entire antibody repertoire. Mutations were called by comparison with the germline sequence. (F) Distributions of the fraction of sequences within each clonal lineage that were the IgM or IgD isotypes among vaccine-responsive and persistent lineages. (G) Fractions of sequences in each clonal lineage that were IgM or IgD, IgG, or IgA. Each dot is a lineage and is positioned according to the isotype composition of that lineage and colored according to identification as vaccine-reponsive (yellow) or persistent (blue).

Fig. 2.

Genetic signatures of somatic evolution in clonal antibody lineages. (A and B) Examples of phylogenies of vaccine-responsive (A) or persistent (B) clonal B cell lineages. Leaves are colored by isotype. Phylogenies are rooted on the germline sequence. (C and D) SFSs averaged across all vaccine-responsive lineages (C) or persistent lineages (D). Error bars indicate SEM. SFSs generated by population genetic models of continuous adaptation driven by strong positive selection (orange), neutral drift with a constant population size (green), and neutral drift with an expanding population (purple) are shown for comparison. Shading indicates SEM across simulations (100 replicates). (E and F) Distribution of significance scores of Fay and Wu’s H statistic for vaccine-responsive lineages or persistent lineages compared against null models of neutral evolution with constant (E) or expanding (F) population size. Distributions for the null models and for simulated populations undergoing continuous adaptation driven by strong selection are also shown. Simulations were performed using population sizes sampled from the observed population-size distributions of vaccine-responsive or persistent lineages (10,000 replicates).

Fig. 3.

Relationship between genetic signatures of selection and clonal expansion after vaccination. (A) Signatures of selection compared with magnitude of clonal expansion after vaccination. Each dot is a lineage. Color indicates whether a lineage has a significant signature of selection, as indicated by the legend in C. (B) Distribution of magnitudes of clonal expansion after vaccination [fold-change (FC) from D0 to D7] among selected and nonselected lineages. Points at “inf” indicate lineages that were detected at D7 but not at D0 and therefore have undefined FC. (C) Fraction of lineages that exhibited clonal expansion with magnitude exceeding various cutoffs among selected and nonselected lineages.

Fig. 4.

Signatures of selective sweeps within multiple subclones of vaccine-responsive antibody lineages. (A) Examples of phylogenies of vaccine-responsive clonal B cell lineages having evidence for selective sweeps favoring multiple subclones. Clades identified as significantly positively selected by our algorithm (P < 0.05) are indicated by arrows and red stars. Leaves are colored by isotype. Phylogenies are rooted on the germline sequence. (B) Distribution of the number of distinct selected subclones (i.e., clades displaying evidence for a selective sweep) within vaccine-responsive lineages having >1,000 sequences. FDR, false discovery rate. (C) Relationship between the number of distinct selected subclones within a clonal lineage and population size (number of sequences) of the lineage. Pearson correlation coefficient is shown.

Fig. 5.

Phylogenetic identification of affinity-enhancing mutations. (A) Example of a phylogeny of a clonal B cell lineage colored by the inferred fitness of each sequence. (B) Regional distribution of nonsynonymous mutations associated with strong fitness enhancements (top three branches ranked by fitness change from parent to child) or diminishments (bottom three branches ranked by fitness change from parent to child), displayed as enrichment relative to synonymous mutations (dN/dS) in the same branches. Dashed line indicates no enrichment. Error bars indicate one SD as determined by bootstrap (100 replicates).