FB5P-seq: FACS-Based 5-Prime End Single-Cell RNA-seq for Integrative Analysis of Transcriptome and Antigen Receptor Repertoire in B and T Cells

Abstract

Single-cell RNA sequencing (scRNA-seq) allows the identification, characterization, and quantification of cell types in a tissue. When focused on B and T cells of the adaptive immune system, scRNA-seq carries the potential to track the clonal lineage of each analyzed cell through the unique rearranged sequence of its antigen receptor (BCR or TCR, respectively) and link it to the functional state inferred from transcriptome analysis. Here we introduce FB5P-seq, a FACS-based 5'-end scRNA-seq method for cost-effective, integrative analysis of transcriptome and paired BCR or TCR repertoire in phenotypically defined B and T cell subsets. We describe in detail the experimental workflow and provide a robust bioinformatics pipeline for computing gene count matrices and reconstructing repertoire sequences from FB5P-seq data. We further present two applications of FB5P-seq for the analysis of human tonsil B cell subsets and peripheral blood antigen-specific CD4 T cells. We believe that our novel integrative scRNA-seq method will be a valuable option to study rare adaptive immune cell subsets in immunology research.

Keywords: B cells; T cells; antigen receptor; single-cell RNA sequencing; transcriptome.

Figure 1

Overview of the FB5P-seq experimental workflow. (A) Major experimental steps of the FB5P-seq workflow. (B) Schematic overview of the molecular designs and the reactions in the FB5P-seq workflow. (C) Schematic illustration of the mapping of the Read1 sequences (in red) on IGH and IGK or IGL amplified cDNA, enabling the in silico reconstruction of paired variable BCR sequences.

Figure 2

Overview of FB5P-seq bioinformatics workflow. Major steps of the bioinformatics pipeline starting from Read1 and Read2 FASTQ files for the generation of single-cell gene expression matrices and BCR or TCR repertoire sequences.

Figure 3

FB5P-seq quality metrics on human tonsil B cell subsets. (A) Experimental workflow for studying the human tonsil B cell subsets with FB5P-seq. (B) Per-cell quantitative accuracy of FB5P-seq computed based on ERCC spike-in mRNA detection (see section Materials and Methods) for memory B cells (Mem, n = 73 tonsil 1, n = 65 tonsil 2), GC B cells (GC, n = 235 tonsil 1, n = 242 tonsil 2), and PB/PCs (n = 78 tonsil 1, n = 152 tonsil 2). The black line indicates the median with 95% confidence interval error bars. (C) Molecular sensitivity of FB5P-seq computed on ERCC spike-in mRNA detection rates (see section Materials and Methods) in two distinct experiments. The dashed lines indicate the number of ERCC molecules required to reach a 50% detection probability. (D,E) Total number of unique genes (D) and molecules (E) detected in human tonsil Mem B cells (n = 73 tonsil 1, n = 65 tonsil 2), GC B cells (n = 235 tonsil 1, n = 242 tonsil 2), and PB/PCs (n = 78 tonsil 1, n = 152 tonsil 2). The black line indicates the median with 95% confidence interval error bars. (F) Gene body coverage analysis for tonsil 1 (n = 5) and tonsil 2 (n = 6) plate libraries. Each curve was computed from the BAM file corresponding to one library from a 96-well plate pool (see section Materials and Methods). (G) Pie charts showing the relative proportion of cells with reconstructed productive IGH and IGK/L sequences (black), only IGK/L sequences (dark gray), only IGH sequences (light gray), or no BCR sequence (white) among Mem B cells, GC B cells, and PB/PCs from tonsil 1 and tonsil 2 samples. The total number of cells analyzed for each subset is indicated at the center of the pie chart.

Figure 4

FB5P-seq analysis of human tonsil B cell subsets. (A) t-SNE map of single human tonsil B cell subsets computed on 4,000 variable genes, excluding the BCR genes. The cells are colored based on the surface protein (upper panel) or the corresponding gene (lower panel) expression of the indicated markers (n = 845 cells). (B) Expression of the indicated marker genes for Mem B cells (upper panel), GC B cells (middle panel), or PB/PCs (bottom panel) in single human tonsil B cells laid out in the t-SNE map. (C) Gene expression heatmap of GC B cells (n = 477), Mem B cells (n = 138), and PB/PCs (n = 230) for the top 10 marker genes of each subset (n.d.: not detected). (D,E) t-SNE map of single human tonsil B cell subsets colored by IGH isotype (NA: not applicable, i.e., no IGH reconstructed) (D) or V_H-J_H repertoire (gray cells: no IGH reconstructed) (E). (F) Scatter plots showing the IGH mutation frequency in human tonsil 1 (circles) and tonsil 2 (triangles) B cells sorted by their IGH isotype and phenotype (Mem B cells: n = 11 IgM/IgD+, n = 37 IgG+, and n = 26 IgA+; GC B cells: n = 55 IgM/IgD+, n = 174 IgG+, and n = 32 IgA+; PB/PCs: n = 4 IgM/IgD+, n = 179 IgG+, and n = 42 IgA+ PB/PCs. The black line indicates the median.

Figure 5

FB5P-seq analysis of the human peripheral blood antigen-specific CD4 T cells. (A) Experimental workflow for studying the human peripheral blood C. albicans-specific CD4 T cells with FB5P-seq. (B) Per-cell quantitative accuracy of FB5P-seq computed based on ERCC spike-in mRNA detection (see section Materials and Methods) for C. albicans-specific CD4 T cells (n = 82). The black line indicates the median with 95% confidence interval error bars. (C) Total number of unique genes detected in C. albicans-specific CD4 T cells (n = 82). The black line indicates the median with 95% confidence interval error bars. (D) Gene expression heatmap of human peripheral blood C. albicans-specific CD4 T cells for a selected panel of marker genes (n.d.: not detected). (E) Pie charts showing the relative proportion of cells with reconstructed productive TCRA and TCRB sequences (black), only TCRB sequences (dark gray), only TCRA sequences (light gray), or no TCR sequence (white) among C. albicans-specific CD4 T cells (n = 82). (F) Distribution of TCRB clones among C. albicans-specific CD4 T cells (n = 67). The black and gray sectors indicate the proportion of TCRB clones (clonotype expressed by at least two cells) within the single cells analyzed (white sector: unique clonotypes). (G) Projection of C. albicans-specific CD4 T cells (n = 82) on the first two PCs computed on 4,000 variable genes, excluding the TCR genes (PC1: 3% of total variability, PC2: 3% of total variability). The cells are colored based on the V_β-J_β repertoire (gray cells: no TCRB reconstructed).