Induction of bronchus-associated lymphoid tissue is an early life adaptation for promoting human B cell immunity

Abstract

Infants and young children are more susceptible to common respiratory pathogens than adults but can fare better against novel pathogens like severe acute respiratory syndrome coronavirus 2. The mechanisms by which infants and young children mount effective immune responses to respiratory pathogens are unknown. Through investigation of lungs and lung-associated lymph nodes from infant and pediatric organ donors aged 0-13 years, we show that bronchus-associated lymphoid tissue (BALT), containing B cell follicles, CD4⁺ T cells and functionally active germinal centers, develop during infancy. BALT structures are prevalent around lung airways during the first 3 years of life, and their numbers decline through childhood coincident with the accumulation of memory T cells. Single-cell profiling and repertoire analysis reveals that early life lung B cells undergo differentiation, somatic hypermutation and immunoglobulin class switching and exhibit a more activated profile than lymph node B cells. Moreover, B cells in the lung and lung-associated lymph nodes generate biased antibody responses to multiple respiratory pathogens compared to circulating antibodies, which are mostly specific for vaccine antigens in the early years of life. Together, our findings provide evidence for BALT as an early life adaptation for mobilizing localized immune protection to the diverse respiratory challenges during this formative life stage.

Extended Data Fig. 1. Germinal Centers in pediatric lungs and associated lymph nodes contain follicular dendritic cells

Tissue sections from lung-associated lymph nodes (LLN) and lungs were stained with anti-CD3 (green), anti-CD20 (blue), anti-CD10 (red), and anti-follicular dendritic cell (FDC) (light blue) antibodies. a, Representative image of LLN from a 1.5 yr old donor showing FDC in a germinal center (GC) structure (scale bar: 150μm) (n=3, with 3 sections imaged per donor). b, FDC in BALT shown in representative images from 4 organ donors paired by row. FDC, and CD10 channels are shown for each donor and merged column shows staining with all four fluorochromes (n=5, with 3 sections imaged per donor). c, LLN from a 4yr old donor showing FDC (white arrows) in GC structures (pink) (scale bar: 200μm). Heterogeneous appearance of FDC with GC B cells across different follicles suggests multiple stages of GC formation (n=3, with 3 sections imaged per donor). FDC=follicular dendritic cell, yr= year, LLN=lung lymph node, GC=germinal center

Extended Data Fig. 2. Gating strategy for analysis of B cell subsets by flow cytometry

Lymphocytes were selected based on forward (FSC) vs side scatter (SSC) height (H) properties, then single cell gating was applied based on SSC-H vs area (−A), followed by selection of CD45+ live cells. B cells (CD19+CD3−) were selected by CD19 expression and lack of CD3 expression and three subpopulations were identified based on IgD and IgM expression: 1. IgD+ cells (second row) were delineated into subsets based on CD27 expression with IgD+CD27− cells further divided based on CD38 and CD10 expression into transitional and naive B cells and IgD+CD27+ cells to IgD+ GC and non-class switched (NCS) memory B cells, which were analyzed for CD69 expression for tissue residency. 2. IgD-IgM− cells were also delineated based on CD27 expression to GC B cells based on CD38 and CD10 expression and CD27+IgD-IgM-CD10− B cells were further delineated into IgA+ memory and IgG+ class-switched (CS) memory B cells which were further analyzed for CD69 expression for tissue residency and CD95 for activation. 3. IgD-IgM+ B cells were delineated into immature subsets based on CD38 and CD10 expression.

Extended Data Fig. 3. Gating strategy for analysis of follicular helper T cells (T_FH) by flow cytometry

Single cells were selected based on side scatter (SSC), height (H) vs area (A) properties, followed by selection of CD45+ live cells. Conventional CD4+ T cells were gated based on lack of CD25 and FoxP3 expression and further gated on CD45RA- non-naïve cells. Tfh subsets were then identified based on expression of CXCR5 and PD-1. GC = germinal center.

Extended Data Fig. 4. Determination of B cell subsets in the LLN from Leiden clustering and BCR clone identification

a, UMAP projection of all B cells colored by original leiden cluster and additional key markers delineating known B cell subsets (bottom rows). b, Dotplot shows top differentially expressed genes (DEGs) per Leiden cluster indicating how clusters were collapsed to delineate known subsets. Gene expression values were scaled to a log2 fold change. Dots are colored by average logFC expression and sized by the percentage of cells per cluster that expressed the particular gene. Gene lists were filtered based on a minimum logFC of >1, adjusted P <0.01 and detected on at least 10% of cells within its cluster. Significant genes calculated using a two-sided Wilcoxon with tie correction. Indicated at the bottom of the dotplot is the Leiden clusters that make up the different B cell subsets based on gene expression and the tissue they are predominantly expressed in. c. Barplots indicating the number of clones for each individual donor in each site, stratified by subsets as identified in (b).

Extended Data Fig. 5. B cells from lung- and gut-associated lymph nodes exhibit similar transcriptional profiles

Single cell RNA-sequencing data derived from 33,737 B cells from the lung-associated lymph node (LLN) and gut-associated mesenteric lymph nodes (MLN) of three donors aged 1–3 years was analyzed as in Fig. 5 and extended data Figure 4. a, UMAP projection of all B cells colored by donor (top), tissue (middle) and subset (bottom), b, Expression of top lineage-defining genes for naïve/transitional (top row), memory (middle row) and germinal center (bottom row) B cells in the UMAP. c, Heat map showing z-score expression of the top differentially expressed genes associated with the major B cell subsets in each row: Transitional (Trans), Naïve, Non-class switched (NCS) memory, Class-switched (CS) memory, and germinal center (GC) B cells. d, Proportional composition of each B cell subset in the LLN and MLN for each donor displayed as a barplot.

Extended Data Fig. 6. Antigen specificity of antibodies derived from lungs, lung-associated lymph nodes, and plasma of individual donors

a, Spider plots showing relative reactivities of IgG and IgA for the different antigen types (green=LLN, pink=lung, violet=plasma) from each pediatric donor (lung/LLN: n=6, plasma: n=6 (3 samples from organ donors and 3 samples from living blood donors) as determined by the antigen array assay (see methods). Tissue and plasma samples derived from the same donors are linked with vertical lines. b, Bar graphs showing geometric mean titers (GMT) of each Influenza strain in serum of adult (red, n=5) and pediatric (green, n=6) donors. Significance was determined by unpaired t test and p value generated by two-tailed test (p ≦0.05). Error bars indicate SD.

Fig. 1:. Pediatric organ donors and their characteristics.

a, Study Schematic: Lungs and lung-associated lymph nodes (LLN) were obtained from deceased pediatric organ donors (n=60) aged 0–13 years of age. Samples were processed as tissue sections for analysis by immunofluorescence imaging, and as single cell suspensions for high-dimensional flow cytometry, BCR sequencing, single-cell RNA sequencing, and assays for antibody-specificity using antigen array analysis. b, c, Donor demographics shown as age distribution, reported sex (female n=18, male n=41), ethnicity/race (African American n=8; blue, Hispanic n=14; red, White n=36; green, American Indian n=1; purple) or cause of death (asphyxiation, n=11; blue, drowning n=12; red, other anoxia n=6; green, head trauma n=24; purple, SIDS n=3; pink, stroke n=3; black, unknown n=1; brown). BCR= B cell receptor, SIDS=Sudden Infant Death Syndrome

Fig. 2:. Formation and regression of BALT containing GC during infancy and childhood.

a, Confocal imaging of lungs from pediatric donors showing immunostaining with antibodies specific for CD20 (green), CD4 (red), CD45RO (blue) and CD4/CD45RO (purple, white arrow) in an infant donor (left), and 1.5 year old donor showing BALT near airways (right). b, Lung imaging as in (a) in donors of indicated ages (white arrows indicate BALT). c, Quantitative analysis of BALT and lymphocyte content over age compiled from 21 donors. Top: Graph shows percent of central airways containing BALT from 2–3 sections per donor (95%CI −0.91 to −0.31, df =19); Middle: Dynamics of B cell and memory CD4 T cell content in lungs over childhood (cells/mm² from 3–5 sections per donor; memory T cells: 95%CI 0.82 to 0.98, df=63 B cells: R2=0.58, df=60); Bottom: Proportion of B cells within BALT over age from 3–5 sections per donor (95%CI −0.71 to 0.29. df=63). Linear regression analysis showing the line of best fit (colored line) and 95% confidence bands (black lines). For B cell dynamics, segmental linear regression was best fitted; dotted lines shows the intersection of two linear regressions (X₀). d, e, GC content in lungs and LLN of a representative 1.5 year donor (d) and donors of indicated ages (e ) shown by staining with anti-CD4 (green), anti-CD20 (blue), and anti-CD10 (red); GC B cells (CD10⁺CD20⁺) appear as pink. f, Dynamics of GC formation in lungs and LLN over childhood shown as proportion of B cell follicles containing GC calculated as mean percentage from 3–5 sections per donor (n=12). Linear regression analysis shows the line of best fit (colored line) and 95% confidence bands (black lines) for GC in LLN (R2=0.64) (95% CI=−0.24 to 0.81, df=10), while a spline was best fit for GC dynamics within BALT. g, AID expression (green) by GC B cells (CD20⁺CD10⁺AID⁺) from 3 representative donors of indicated ages. h. Frequency of AID⁺ B cells (2–3 sections per donor) within GC over age (n=12). Segmental linear regression was best fit. (R2=0.688, df=20). *r value was generated by Pearson’s correlation analysis and significance determined by two-tailed test. BALT=Bronchus Associated Lymphoid Tissue, LLN=lung lymph node, CI=confidence interval, df=degree of freedom, GC=germinal center, AID=activation-induced cytidine deaminase

Fig. 3:. Distinct B cell subset composition and age-associated dynamics in lungs and LLN.

High-dimensional flow cytometry was performed on B cells isolated from lungs and LLN from donors 0–13 years of age (n=35). a, Schematic shows the major B cell subsets, their maturation state, and distinguishing markers. b, Distinct proportion of B cells (among CD45⁺ cells) and composition of B cell subsets (among CD19⁺cells) in lung and LLN compiled from 35 donors. Significance determined by two-tailed T test (significance ≦0.05). Error bars indicate SD. c, Differential expression of markers for follicular subsets (CXCR5), activation (CD95), and tissue residence (CD69) by indicated B cell subsets in lung and LLN shown in representative flow plots from individual donors (top row), and boxplots of compiled data from 35 donors (lower row). Midline of boxplots in (c) represents the median, the box is the interquartile range and the whiskers are the minimum and maximum values. Significance determined by two-tailed T test (significance ≦0.05). d, Distinct age-associated changes for B cell subsets in lungs (first and third column, n=29) and LLN (second and fourth column, n=30) over the first 12 years of life. Linear regression analysis showing the line of best fit (solid line) and 95% confidence bands (dotted lines). r values were generated by Pearson’s correlation analysis and significance determined by two-tailed test. (Total B Lung: 95% CI −0.66 to −0.02, df=28. Total B LLN: 95% CI 0.12 to 0.70, df=29. Transitional B Lung: 95% CI −0.85 to −0.46, df=28. Transitional B LLN: 95% CI −0.81 to −0.36, df=29. Naive B Lung: 95% CI −0.43 to 0.28, df=28. Naive B LLN: 95% CI −0.50 to 0.19, df=29. GC B Lung: 95% CI −0.53 to 0.16, df=28. GC B LLN: 95% CI −0.53 to 0.14, df=29. IgG memory Lung: 95% CI 0.53 to 0.87, df=28. IgG memory LLN: 95% CI 0.12 to 0.69, df=29. IgA memory Lung: 95% CI 0.59 to 0.89, df=28. IgA memory LLN: 95% CI 0.29 to 0.78, df=29). LLN=lung lymph node, GC=germinal center, CS=class-switched, Mem=memory, NCS=non-class-switched

Fig. 4:. Analysis of T –helper subsets in lungs and associated lymph nodes.

T-follicular helper (T_FH) cell content in lungs and Lung-associated LN (LLN). a, Representative flow cytometry plots from lung and LLN of a 10yr old donor delineating subsets of non-naive (CD45RA⁻) CD4⁺T cells based on PD-1 and CXCR5 expression (left). Boxplots indicating the frequencies of the indicated subsets in the lung (top) and LLN (bottom) compiled from 19 donors (PD-1^lo, green; PD-1^intCXCR5⁺, black; PD-1^hiCXCR5⁺, pink; PD-1^hiCXCR5⁺, purple). b, Bcl-6 expression within CD4⁺T cells subsets in the lung (top, stratified into PD-1^lo (green) vs PD-1^hi (light blue)) and LLN (bottom, stratified based on PD-1 and CXCR5 expression as in (a)) shown in representative histograms (left) and boxplots of Bcl-6 MFI (right). c, ICOS and TIGIT expression by subsets delineated in (b) from the lung (top) and LLN (bottom) shown in representative flow cytometry plots (left) and boxplots of the percent ICOS⁺TIGIT⁺ cells (right). d, Bcl-6 expression by PD-1 subsets differentially expressing ICOS and TIGIT in the lung (top) and LLN (bottom) show as representative histograms (left) and boxplots of Bcl-6 MFI (right). Midline of boxplots in a-d represents the median, the box is the interquartile range and the whiskers are the minimum and maximum values. Significance for b-d was determined by a two-tailed T test (if two groups) or One-way ANOVA with Tukey’s multiple comparisons test (if three or more groups) (significance ≦0.05). e, Diagram showing distinct phenotypes of T_FH-like cells in LN and lungs. f, Frequency of PD-1^hiICOS⁺TIGIT⁺ cells in the lung and LLN over age. Linear regression analysis showing the line of best fit (colored line) and 95% confidence bands (black lines); r value was generated by Pearson’s correlation analysis and p value was generated by two-tailed test. (Lung: 95%CI −1.14 to −0.14, df=18, LLN: 95%CI −2.1 to 1.4, df=16). T_FH=follicular helper T cell, GC=germinal center, CI=confidence interval, df=degree of freedom, ns=non-significant

Fig. 5:. Transcriptional signatures and clonal dynamics for lung and LLN B cells in early life.

Single cell transcriptome and BCR sequencing was conducted on 26,745 B cells from the lung and LLN of three donors aged 1.33, 2 and 3yrs. a, UMAP of all B cells colored by donor, tissue and subset (trans – transitional, NCS – non-class-switched memory, GC – germinal center, CS – class-switched memory), or shaded by expression of key markers delineating subsets. b, Heatmap showing differentially expressed genes and their z-score expression within each subset across sites. c, Barplot showing B cell subset proportions in each site of individual donors. d, Volcano plots showing differentially expressed genes between the lung and LLN when comparing combined naive and transitional B cells (top) and memory subsets (bottom). Red dots indicate genes upregulated in LLN and blue dots those upregulated in the lung (minimum log2FC >1, maximum P value <0.01). Significance calculated using a two-sided Wilcoxon with tie correction. e, Heatmap showing the z-score expression of the consensus genes between B cell subsets in lung compared to LLN. f, UMAP (left) of the enrichment score of genes associated with B cell activation (see methods); violin plot showing the distribution of enrichment score of these genes within each subset for each organ (right). g, BCR analysis showing the fraction of clones with ≧2 cells per donor per tissue. h, Bar plot indicating the fraction of mutated clones (those with an average ≧2% mutated positions in a sequence) per donor by site (lung, blue; LLN, red). i, Violin plots indicating average mutations per clone for each subset and tissue (lung, blue; LLN, red) within each donor. Horizontal lines indicate interquartile range. j, Paired dotplot indicating the median distribution of the ratios of the average number of non-synonymous (R) and synonymous (S) mutations per clone within the framework (FWR) (circles) and complementary determining regions (CDR) (triangles) of the BCR, displayed by tissue and subset (GC and CS) in log scale. Statistical significance determined by multiple T-tests using the Holm-Sidak method. Adjusted P values: lung GC = 0.001341, lung CS = 0.000081, LLN GC = 0.323825, LLN CS = 0.003012.

Fig. 6:. B cell clonal expansion and somatic hypermutation in the lung and LLN during early life and childhood.

B cell receptor (BCR) VH chains were sequenced from the lung and LLN of six pediatric donors aged 7w-6y and three reference adult donors. a, Proportion of top n B cell clones (ranked by number of unique sequences) per total unique sequences from lung and LLN. b, Clone numbers detected in the lung (blue) and LLN (red) shown in Venn diagrams as unique or shared between sites. c, Bar graph showing fraction of clones found only in lung (blue, left) or LLN (red, right) that are mutated (average per clone of fraction of mutated positions in a sequence ≧0.02) in each donor. d, Fraction of shared clones that are unmutated in both sites (average per clone of fraction of mutated positions in sequence <0.02) (white) or mutated (average per clone of fraction of mutated positions in a sequence ≧0.02) in both tissues (green), only in the LLN (red), or only in the lung (blue). e, f, Range of mutation frequencies shown in violin plots and scatter plots of shared clones mutated in both sites (e) or one site only (f). (Blue: distribution in the lung; red: distribution in the LLN). LLN=lung lymph node, w=weeks, m=months, y=years.

Fig. 7:. Distinct antigen specificities for tissue B cells compared to circulating antibodies.

B cells were isolated from the lung and LLN of donors 1–3 yrs of age, activated in vitro and secreted antibodies were analyzed for specificity using antigen arrays (see methods) along with paired and age-matched plasma samples. a, Spider plots showing relative IgG (left) and IgA (right) reactivities for the different antigen types (green=LLN, pink=lung, top row; blue=plasma, bottom row) from pediatric donors (average of three donors for lung/LLN and average of six donors for plasma). b, Bar graph showing geometric mean titers (GMT) of each Influenza strain for lung (blue), LLN (red), and plasma (green). Each circle represents an individual donor (lung/LLN: n=3, plasma: n=6). Significance determined by One-way ANOVA and Tukey’s multiple comparisons test. Error bars indicate SD. c, B cell activation in the lung during respiratory infection. Graphs show indicated B cell subset frequency in the lung from three donors (0–4 years old) with acute respiratory infection and age-matched control donors (n=14). Significance determined by unpaired T test and p value generated by two-tailed test (significance ≦0.05). LLN=lung lymph node, Mem=memory, CS=class-switched, ns=non-significant