Fig. 5: Integrated analysis reveals two compatible development models for memory B cells.

Transcriptomic-based inference of developmental trajectory for three memory B cell subpopulations with monocle2. Both B cell subpopulation distribution (A) and pseudotime prediction (B) are shown along the trajectory. C IGHG and IGHA subclass composition for three memory B cell subpopulations. Numbers in the parentheses indicate the number of B cells. D Doughnut plot showing the tissue-specific composition of three memory subpopulations for the mixed samples and three donors. E CDR3 length distribution. The paired smooth curves are distributions fitted from kernel density estimates. Classical, IgM⁺ and CD27^-IgM⁺IgD⁺ subpopulations are denoted as “C”, “M” and “MD”, respectively. F, G BCR heavy chain variable (H) and joining (I) gene usage for classical, IgM⁺ and CD27⁻IgM⁺IgD⁺ memory B cell subpopulations. For variable genes, only those with a usage frequency of at least 1% for classical memory B cells are shown. The error bar shows a 95% confidence interval. H Isotype-specific comparison of SHM level of heavy chain variable genes between three memory subpopulations. Numbers on the boxes indicate the number of unique clonotypes. I Venn diagram showing clonotype sharing between classical, IgM⁺ and CD27⁻IgM⁺IgD⁺ memory B cell subpopulations. The numbers with a pink background indicate the clonotype similarity as measured by the Jaccard similarity index. J Barplots showing the fraction of clonotype sharing between donors, sources and isotypes. Statistical test used in (D–G) is two independent samples t test, two-sided. *, p < 0.05; **, p < 0.01; ***, p < 0.001.