Frequency and mutational status of antibody isotypes in healthy and in leukemic repertoires. (A) Frequency of unique B-cell receptor (BCR) corresponding to each isotype across total CLL repertoires (red), healthy peripheral blood mononuclear cell repertoires (blue), and non-malignant CLL repertoires (excluding the major malignant clones) (green). Frequency of each isotype (sub)class is represented as a percentage of the unique BCRs corresponding to a specific isotype per individual. (B) Differences in complementarity determining region 3 (CDR3) amino acid lengths of unique BCRs corresponding to each isotype calculated as mean CDR3 length per isotype per sample. (C) Differences in non-silent mutations in IGHV genes of unique BCRs calculated as median non-silent mutation per isotype per sample. (D) Differences in silent mutations in IGHV genes of unique BCRs calculated as median silent mutation per isotype per sample. In all panels ***p-value < 0.0005; **p-value < 0.005; and *p-value < 0.05 (Wilcoxon rank-sum test of the respective metric in each panel compared between healthy and disease groups).

Relationship between somatic hypermutation and class-switching in the evolution of B-cell responses in healthy repertoires. Repertoires were clustered into networks of related B-cell receptors (BCRs) as previously described (). (A) Boxplots of the mean number of mutations from germline in clones exhibiting 2 isotype classes or greater than 2 isotype classes that are either IgM⁺IgD⁺ or IgM⁻IgD⁻. The y-axis represents the number of nucleotide mutations in IGHV genes of BCRs associated with the respective number of isotypes. (B) Boxplots of the mean size of BCR clones formed by the respective number of isotypes shown on the x-axis as a proportion of total BCR reads. (C) Boxplots of the mean size of clones exhibiting 2 isotype classes that are either IgM⁺IgD⁺ or IgM⁻IgD⁻. Numbers above each boxplot in (A–C) represent the total number of clones considered for each isotype group. (D) A maximum parsimony tree of a class-switched BCR cluster in healthy repertoire representing an example of a clonal expansion following class-switching. ***p-value = 0.0005 and **p-value = 0.005 (Wilcoxon rank-sum test).

Relationship between somatic hypermutation, class-switching, and clonal expansion in the evolution of malignant B-cell clones. (A) Isotype composition of the largest CLL clone across all leukemic individuals. Isotype frequency was calculated as the percentage of reads in the CLL clone corresponding to a given isotype, on a square-root scale to distinguish between low frequency isotypes. (B) Number of mutated nucleotides in IGHV genes of B-cell receptors (BCRs) of different isotypes compared to the IGHV gene of the major leukemic CLL clone for each sample. (C) Unrooted phylogenetic tree of an example class-switched CLL clone. Each tree tip represents a unique BCR sequence, the pie charts represent the proportion of isotypes per unique BCR, and size corresponds to the proportion of BCRs represented by that sequence (square-root scale for visualization). (D) Heatmap of isotype frequency in total CLL repertoires of the six leukemic patients across timepoints of longitudinal sampling (see Table S4 in Supplementary Material). (E) Relationship between degree of class-switching and CLL clone size; the x-axis represents the total frequency of switched BCRs in the malignant CLL clone for each individual (represented as a dot). The y-axis reflects the size of the major CLL clone per individual shown as number of BCRs per individual contributing to the largest CLL clone (as a percentage of the total BCR repertoire). The correlation coefficient labels represent the correlations between percentage switched isotype per largest CLL clone and size of the largest clone for the two sampling points. (F) Frequency of switched BCR in the CLL malignant clones of each patient calculated for the two sampling points (T1 and T2) per sample. For sampling point details, see Table S4 in Supplementary Material.

Sequential class-switching and isotype co-evolution in healthy and in malignant B-cell clones. (A) Boxplots of the proportion of observed class-switch events [B-cell receptor (BCRs) with identical IGHV-D-J regions but different isotypes] between isotypes for healthy individuals and CLL patients. This was calculated as the number of unique BCRs shared between any two isotypes after accounting for read depth through subsampling (see ), normalized to the total number of events. (B) Relative proportional frequencies of different isotype classes in healthy repertoires and in malignant clones across CLL patients. Malignant clones were derived for each CLL sample by removing the non-malignant component of the repertoire (i.e., the BCRs with VJ genes and complementarity determining region 3 amino acid regions different from the most frequent BCR per sample) from the total CLL repertoire. (C) Representative networks of the relative class-switch event frequencies between pairs of isotypes calculated as described in (A). The thickness of the lines represents the mean relative class-switch event frequencies calculated as in (A), and the green and light blue lines connect isotypes that exhibit significantly higher (green) or lower (light blue) overlap probability in CLL repertoires. The gray lines in the “CLL” panel represent combinations of isotypes with non-significant (p-value > 0.05) difference in the frequency of class-switch events compared to healthy repertoires.

Role of somatic hypermutation and variable gene identity in the class-switch fate of healthy B-cell clones. (A) Differences in IGHV (upper panel) and IGHJ (lower panel) gene family usages between different isotype classes in healthy individuals. Frequencies of individual IGHV-J gene families are calculated as percentage of unique B-cell receptor (BCR) reads from each IGHV or IGHJ gene family associated with the respective isotype. (B) Heatmap of IGHV gene frequencies across isotypes. Individual values are scaled by row, key represents row-Z score and distance matrix between IGHV genes was generated using Euclidian distance. Only IGHV genes with >100 reads were considered. (C) Correlation of IGHV gene frequencies across isotypes. Only unique BCR sequences were considered and IGHV genes represented by less than 100 reads were excluded. The size and the color of the squares represent the absolute value and the direction of the correlation, respectively. The values in the colored squares represent the p-value of the significant positive or negative correlations across isotypes. Blank squares represent non-significant correlations (p-value > 0.05). The scatterplot matrix represents the frequencies of individual IGHV genes in the respective isotypes. Trendlines are shown in red. (D) Conditional class-switch probabilities of BCRs given in their IGHV gene identity and mutational status. Boxplots represent the variance of conditional probabilities of individual IGHV genes from the respective IGHV gene families. Only IGHV genes with >100 reads in both mutated and unmutated state were considered. ***p-value < 0.0005; **p-value < 0.001; and *p-value < 0.05 (Wilcoxon rank-sum test).