Genetic measurement of memory B-cell recall using antibody repertoire sequencing

Abstract

Annual influenza vaccinations aim to protect against seasonal infections, and vaccine strain compositions are updated every year. This protection is based on antibodies that are produced by either newly activated or memory B cells recalled from previous encounters with influenza vaccination or infection. The extent to which the B-cell repertoire responds to vaccination and recalls antibodies has so far not been analyzed at a genetic level--which is to say, at the level of antibody sequences. Here, we developed a consensus read sequencing approach that incorporates unique barcode labels on each starting RNA molecule. These labels allow one to combine multiple sequencing reads covering the same RNA molecule to reduce the error rate to a desired level, and they also enable accurate quantification of RNA and isotype levels. We validated this approach and analyzed the differential response of the antibody repertoire to live-attenuated or trivalent-inactivated influenza vaccination. Additionally, we analyzed the antibody repertoire in response to repeated yearly vaccinations with trivalent-inactivated influenza vaccination. We found antibody sequences that were present in both years, providing a direct genetic measurement of B-cell recall.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

Consensus read approach validation. (A) Experimental setup. Blood is drawn from an individual. Two separate PBMC aliquots are prepared, and RNA is extracted. Several sequencing libraries are prepared and sequenced. (B) Venn diagrams illustrating the IGH sequence overlap between L1 D7 and Lib. Rep. for all IGH sequences or only abundant IGH sequences represented by more than 5 IGH molecules. (C) The abundances of IGH sequences shared between L1 D7 and Lib. Rep. are shown as a scatterplot. (D–I) IGH sequences in L1 D7 are ordered by abundance. (D) Abundance (IGH molecules/IGH sequence) of sequence. In bins of 300 sequences from left to right, percents of shared sequences between L1 D7 and indicated samples [(E) Seq. Rep., (F) Lib. Rep., and (G) Bio. Rep.] are shown. (H) Isotype distribution of sequences. (I) Mutation rate (%) of L1 D7 sequences separated by isotypes (IgM, IgG, and IgA).

Fig. 2.

Repertoire size estimation. (A) Visualization of lineages shared between replicate sample pairs (R1 and R2) from individuals (L1, T6, T7, and T8). Data were subsampled to 50,000 IGH molecules to normalize for variability in sampling depth. Lineages of each time point are plotted on the circumference of the circle, with the gray area representing abundance of the respective lineages (logarithmic). Lineages present in two time points are connected with lines colored according to their isotype. (B) Capture–recapture estimation of different IGH lineage populations. Average (±SD) of four individuals is shown.

Fig. 3.

Influenza vaccination response. (A) Change in the number of abundant lineages (containing at least 5 IGH molecules) in response to vaccination by LAIV or TIV is shown for 14 individuals each. Data were subsampled to 50,000 molecules to normalize for variability in sampling depth. Data are shown for each isotype separately. P value is determined by a Mann–Whitney u test. (B) IGH repertoire at days 0, 7, and 28 for individual T4. Each sample was subsampled to 50,000 IGH molecules to normalize for variability in sampling depth. Sequences were ordered by abundance, and each sequence is shown as a vertical bar [the height depicts its abundance (logarithmic)]. The color of each bar indicates the isotype of the sequence.

Fig. 4.

TIV recall response. (A) Number of lineages shared between indicated time points normalized to the overall amount of lineages. Average of five individuals is shown as heatmaps for IgG, IgA, and IgM. (B) Median abundance of IgG, IgA, and IgM lineages shared between the indicated time points.

Fig. 5.

Lineage recall. (A and B) Visualization of lineages shared between time points. Data were subsampled to 50,000 IGH molecules for each time point to normalize for variability in sampling depth. Lineages of each time point are plotted on circumference of the circle, with the gray area representing the abundance of the respective lineages (logarithmic). Lineages present in two time points are connected with lines colored according to their isotype. All shared lineages are shown in A, and shared IgG, IgA, and IgM lineages are shown separately in B. (C) Shared IgG lineages between Y1 D7 and Y2 D7 time points of three individuals are shown as in A.