Maturation of germinal center B cells after influenza virus vaccination in humans

Abstract

Germinal centers (GC) are microanatomical lymphoid structures where affinity-matured memory B cells and long-lived bone marrow plasma cells are primarily generated. It is unclear how the maturation of B cells within the GC impacts the breadth and durability of B cell responses to influenza vaccination in humans. We used fine needle aspiration of draining lymph nodes to longitudinally track antigen-specific GC B cell responses to seasonal influenza vaccination. Antigen-specific GC B cells persisted for at least 13 wk after vaccination in two out of seven individuals. Monoclonal antibodies (mAbs) derived from persisting GC B cell clones exhibit enhanced binding affinity and breadth to influenza hemagglutinin (HA) antigens compared with related GC clonotypes isolated earlier in the response. Structural studies of early and late GC-derived mAbs from one clonal lineage in complex with H1 and H5 HAs revealed an altered binding footprint. Our study shows that inducing sustained GC reactions after influenza vaccination in humans supports the maturation of responding B cells.

Figure 1.

Persistence of vaccine-specific GC B cells after human influenza virus vaccination. (A) Schematic of study design. Eight healthy adults (aged 26–40) received the 2018 QIV intramuscularly. Blood, FNAs of the ipsilateral axillary lymph nodes (LN), and bone marrow (BM) were collected prior to vaccination (week 0) and at the indicated weeks after vaccination. (B) Kinetics of HA-binding PBs (CD20^lo HA⁺) in blood from seven participants. (C) 2018 QIV–specific IgG plasma antibody titers were measured via ELISA for seven participants. (D) Representative flow cytometry gating of total GC B cells (CD20⁺ CD38^int) and HA-binding GC B cells (CD20⁺ CD38^int HA⁺) in the LN from participant 05. Cells were pregated on CD4⁻ CD19⁺ IgD⁻ lymphocytes. (E) Kinetics of total GC B cells (open circles) and HA-binding GC B cells (closed circles) for all participants as defined by gating in D. Daggers indicate samples were excluded due to low cell recovery or blood contamination. (F) Representative ELISpot wells coated with 2018 QIV or anti-immunoglobulin (Ig) and developed in blue (IgG) and red (IgA) after plating BMPCs from participants 04, 05, and 11. (G) Frequencies of IgG and IgA 2018 QIV–specific BMPCs measured by ELISpot for seven participants. Participants with a detectable HA-binding GC are colored light blue. LoD, limit of detection. See also Fig. S1.

Figure S1.

Persistence of vaccine-specific GC B cells after human influenza virus vaccination. Related to Fig. 1. (A) Flow cytometry gating strategy for PBs from blood. (B) Flow cytometry gating strategy for HA-binding GCs in lymph nodes (LN). (C) Flow cytometry gating of total GC B cells (CD20⁺ CD38^int) and HA-binding GC B cells (CD20⁺ CD38^int HA⁺) in the LN from participant 04. Cells were pregated on CD4⁻ CD19⁺ IgD^lo live singlets. Dagger indicates samples were excluded from further analysis due to low cell recovery or blood contamination.

Figure S2.

Tracking vaccine-specific B cells in persistent GCs. Related to Fig. 2. (A) Unsupervised clustering visualized via UMAP based on scRNA-seq gene expression of all cells pooled from all blood, lymph node (LN), and bone marrow (BM) samples and time points from participants 04, 05, and 11. (B) Dot plot of the average log-normalized expression of marker genes and the fraction of cells expressing the genes in each cluster from A. (C) Annotated UMAP clusters of scRNA-seq samples pooled from blood, LN, and BM samples from all time points from participants 04, 05, and 11. (D) Dot plot for annotated clusters in C. (E) Unsupervised clustering visualized via UMAP based on scRNA-seq gene expression of cells in the B cell cluster from A, pooled from all blood, LN, and BM samples and time points from participants 04, 05, and 11. (F) Dot plot of the average log-normalized expression of marker genes and the fraction of cells expressing the genes in each cluster from E. (G) Annotated UMAP clusters of cells from the B cell cluster in E, pooled from all blood, LN, and BM samples from all time points from participants 04, 05, and 11. (H) Dot plot for annotated clusters in G. (I) Optical density (OD) at 490 nm as determined by ELISA of 2018 QIV–binding clonally unique mAbs generated from GC B cells from week 13 and week 17 from participants 05 and 04, respectively. Positive binding was defined as greater than two times the OD 490 value for antibody binding to BSA. (J) Frequency of 2018 QIV–specific GC B cell clones at the indicated time points in participant 04. Each slice represents one clonal family. The frequency of a clonal family is defined as the percentage of cells in each clonal family among the total GC B cells at each time point (n = 166 at week 1, n = 338 at week 2, n = 930 at week 17). Colored slices indicate clones identified at multiple time points.

Figure 2.

Tracking vaccine-specific B cells in persistent GCs. (A and B) Clustering visualized via UMAP of B cells from blood, lymph node (LN), and bone marrow (BM) scRNA-seq samples in participant 05 (A) and participant 04 (B). Each dot represents a cell, colored by phenotype as defined by transcriptomic profiles. Naïve B cells (gold), PBs (red), ABCs (green), GC B cells (blue), LNPCs (red), RMBs (lavender), and plasma cells (PC, red) populations are pooled from all time points (first panel). 2018 QIV–specific cells at each week after vaccination are colored as described. (C) Frequency of 2018 QIV–specific GC B cell clones at the indicated time points in participant 05. Each slice represents one clonal family. The frequency of a clonal family is defined as the percentage of cells in each clonal family among the total GC B cells at each time point (n = 206 at week 2, n = 149 at week 4, n = 293 at week 9, n = 273 at week 13). Colored slices indicate clones identified at multiple time points. N.D., no data. See also Fig. S2; and Tables S3, S4, S5, S6, and S7.

Figure 3.

Affinity-matured late GC B cells bind and neutralize heterologous influenza viruses. (A) Lineage tree of clonal family 05.89107.H1N1 from participant 05. IGHV and IGHJ gene use is indicated. Cells from which mAbs are derived are labeled with the cell of origin and the week isolated. Scale bar indicates mutations per codon. P value is calculated as described in the methods. (B) Optical density (OD) at 490 nm of A/Michigan/45/2015 H1-binding mAbs from clonal family 05.89107.H1N1. (C) Binding of Fabs from clonal family 05.89107.H1N1 to A/Michigan/45/2015 H1 as measured by BLI. K_D, dissociation constant. (D) Binding of mAbs from clonal family 05.89107.H1N1 to the indicated H1, H2, and H5 HA proteins as measured by ELISA. Scale bar is the area under the curve (AUC). (E) Minimum concentration of antibody required to neutralize the indicated viruses. CR9114 was used as a positive control. Each point represents one of two replicates. See also Fig. S3.

Figure S3.

Affinity-matured late GC B cells can bind and neutralize heterologous influenza viruses. Related to Fig. 3. (A) Lineage tree with corresponding heavy chain amino acid sequences in clonal family 05.89107.H1N1. Each row corresponds to its aligned node. Each row corresponds to the adjacent tip of the tree. (B) Lineage tree of clonal families 05.111394.H3N2, 05.129238.B/Ph, and 05.173963.H3N2 from participant 05 and 04.113954.B/Ph and 04.46156.B/Ph from participant 04. IGHV and IGHJ gene use is indicated. Cells from which mAbs are derived are labeled with the cell of origin and the week isolated. P values are calculated as described in the methods. (C) Optical density (OD) at 490 nm of A/Singapore/INFIMH-16-0019/2016 H3-binding or B/Phuket/3073/2013 HA-binding mAbs from clonal families in B. (D) Binding of Fabs from clonal family 05.111394.H3N2 to A/Singapore/INFIMH-16-0019/2016 H3 and Fabs from clonal family 04.113954.B to B/Phuket/3073/2013 HA as measured by BLI. (E) Median half-maximal binding concentration of mAbs isolated from GC B cell clonal lineages from participant 04 (purple) and participant 05 (blue) early (week 2/4) or late (week 9/13/17) after vaccination as measured by ELISA to relevant HA proteins. Filled circles indicate the clonal lineage tree is included in the manuscript. Median values are indicated above each column. P value was determined by paired t test. K_D, dissociation constant.

Figure S4.

Structural comparison of 05.GC.w2.3C10-H1, 05.GC.w13.01-H1, 05.GC.w13.02-H1, and 05.GC.w13.02-H5. Related to Figs. 4 and 5. (A) Overall binding model of each HA-Fab complex from their x-ray (panel 1, 3, 4, and 5) and cryo-EM (panel 2) structures. The Fabs are shown in surface representation. HA is represented as a trimer in backbone cartoon, with one HA monomer highlighted to illustrate the interaction between HA and Fab. (B and C) Local resolution (B) and Fourier Shell Correlation (C) of the cryo-EM structure of 3C10 Fab in complex with A/Solomon Islands/3/2006 (SI06) H1N1 HA. (D) The CDR loops of each Fab in contact with HA are shown with the loops as cartoons. (E) Hydrophobic residues involved in the contact between Fabs and HA are represented as sticks. Three hydrophobic areas in the stem region in H1 HA are circled and the underlying surface is represented in a green hydrophobicity gradient calculated by Color (https://pymolwiki.org/index.php/Color). (F) Critical hydrophobic residues in the heavy chain involved in interaction between Fab and HA are shown in side-chain sticks. (G) Superimposition of each Fab-HA complex and Apo-H1. The Thr61 side chain is illustrated as an indicator of the relative disposition of the interhelical loop among these structures. Distances in Angstroms were measured from T61 in 05.GC.w13.02-H1 to Apo-H1, and between 05.GC.w13.02-H1 and 05.GC.w13.02-H5. 05.GC.w2.3C10 is in yellow, 05.GC.w13.01-H1 in sand, 05.GC.w13.02-H1 in teal, 05.GC.w13.02-H5 in orange, and Apo-H1 in light blue (PDB: 4M4Y). The HA is shown in a backbone cartoon. (H) The extent of the upper pocket in each complex is measured from T290 (S290 in 05.GC.w13.02-H5) in the 290-loop to T61 in the interhelical loop after superimposition of HA1 and HA2 from each complex and Apo-H1. (I) The flexible 290-loop, 300-loop, and interhelical loop in the upper pocket in the stem region are highlighted in red. (J) 05.GC.w2.3C10-H1_SI06 complex determined by cryo-EM is shown in green. The extent of the upper pocket is measured from the distance between S290 and T61 in the 290-loop and interhelical loop, respectively (left). Superimposition of 05.GC.w2.3C10-H1_SI06 complex onto the 05.GC.w2.3C10-H1 (right). (K) Binding mode of Fab HCDR1, HCDR2, and HCDR3 loops with H1 HA. Residues involved in hydrophobic interactions are in red with black labels for HA residues and CDR loops. (L) Binding orientation of the Fab VH domain to HA and location of residues 98 in 05.GC.w2.3C10, 05.GC.w13.01, and 05.GC.w13.02 compared to CR9114. (M) Binding kinetics of 05.GC.w13.02 with alanine substitutions in critical contact positions using BLI. Black lines illustrate the response curves representing a 1:1 binding model of Fab with CA04 H1 HA.

Figure 4.

Crystal structure of 05.GC.w13.01 and 05.GC.w13.02 in complex with A/California/04/2009 (CA04) H1N1 HA and A/Vietnam/1203/04 (Viet04) H5N1 HA. The CA04 H1N1 HA is shown as a molecular surface in white for HA1 and gray for HA2. Residues involved in side-chain and backbone interactions between HA and Fab are represented in white (HA1) and gray (HA2) sticks. HCDR and LCDR represent CDRs in the heavy (H) and light (L) chains. Hydrogen bonds and salt bridges are indicated with black dashes. Fab residues are in Kabat numbering throughout. 05.GC.w13.01-H1 is in sand, 05.GC.w13.02-H1 in teal, 05.GC.w13.02-H5 in orange, and CR9114 in maroon. Fabs and HAs are shown in backbone cartoons with interacting side chains in sticks. For clarity, HA and Fab residues are colored with black and red numbers, respectively. (A) Structural comparison of HFR1-HCDR1 in 05.GC.w13.01-H1, 05.GC.w13.02-H1, and 05.GC.w13.02-H5. (B) Sequence comparisons of the IGHV1-69*12 germline, 05.GC.w2.3C10, 05.GC.w13.01, 05.GC.w13.02, and CR9114. Yellow highlights two critical motifs in HFR1 and HCDR2 in each Fab. (C) Interactions between the TF motif and the upper pocket of HA in each complex are shown. Residues involved in the interactions within this pocket are depicted in sticks. (D) The critical motif for residues 53 and 54 in HCDR2 for each Fab is shown: IV in VF in 05.GC.w2.3C10, 05.GC.w13.01, IL in 05.GC.w13.02, and IF in CR9114. This motif binds to the lower pocket in the HA stem. See also Fig. S4 and Tables S1 and S2.

Figure 5.

Characterization of 05.GC.w2.3C10 and 05.GC.w13.02 in complex with A/California/04/2009 (CA04) H1N1 HA. 05.GC.w2.3C10 and 05.GC.w13.02 are in yellow and teal, respectively. For clarity, HA and Fab residues are colored with black and red numbers, respectively. (A) Structural comparison of the binding of HFR1-HCDR1 with 05.GC.w2.3C10 and 05.GC.w13.02. The red circles indicate the TF motif in 05.GC.w2.3C10 that is lower than the corresponding motif in 05.GC.w13.02. (B) Detailed molecular interactions between 05.GC.w13.02 (left) and 05.GC.w2.3C10 (right) with CA04 HA. (C) The role of the hydrophobic HFR3 Leu74 in 05.GC.w13.02 (left), and the different binding of Ser74 in 05.GC.w2.3C10 (middle). The right panel shows that F29 from the TS motif in 05.GC.w2.3C10 occupies the middle hydrophobic groove. (D) Binding affinity of 05.GC.w13.02 with deletions of residues 31 and 32 to H1 (A/Michigan/45/2015) using BLI. See also Fig. S4 and Tables S1 and S2.

Figure 6.

Re-immunization of individuals with persistent GCs promotes the development of affinity-matured PBs. (A) Schematic of study design. Three participants (04, 05, 11) received the 2019 QIV intramuscularly. Blood, FNAs of the ipsilateral axillary lymph nodes (LN), and bone marrow (BM) were collected prior to vaccination (week 0) and at the indicated weeks after vaccination. (B) ELISpot quantification of 2019 QIV–binding IgG-secreting PBs in blood at weeks 0, 1, and 2 after vaccination for three participants. (C) Median IGHV gene mutation frequency of QIV-binding PB clonal groups found after vaccination with 2018 and 2019 QIV. Lines between points indicate the clone was identified in the PB compartment after both vaccinations. Median IGHV nucleotide mutation frequency of all clonal groups is indicated above each column and the number of clones (n) is indicated in the bottom right of each panel (04, n = 73; 05, n = 122; 11, n = 65). P values were determined by paired t test. (D) Clustering was visualized via UMAP of B cells from blood, LN, and BM scRNA-seq samples in participant 05. Each dot represents a cell, colored by phenotype as defined by transcriptomic profiles. Naïve B cells (gold), PBs (red), ABCs (green), GC B cells (blue), LNPCs (red), RMBs (lavender), and plasma cells (PC, red) populations are pooled from all time points (first panel). QIV-specific cells at each week after vaccination are colored as described. (E) Median IGHV gene mutation frequency of QIV-binding GC B cell clonal groups found after vaccination with 2018 and 2019 QIV. Lines between points indicate the clone was identified in the GC B cell compartment after both vaccinations. Median IGHV nucleotide mutation frequency of all clonal groups is indicated above each column and the number of clones (n = 135) is indicated in the bottom right. P values were determined by paired t test. N.D., no data. See also Fig. S5 and Tables S3, S4, S5, S6, and S7.

Figure S5.

Re-immunization of individuals with persistent GCs engages affinity-matured PBs. Related to Fig. 6. (A) Kinetics of HA-binding PBs (CD20^lo HA⁺) in blood from three participants (04, 05, 11) by flow cytometry. Cells were pre-gated on CD4⁻ CD19⁺ IgD⁻ live singlets. (B) Optical density (OD) at 490 nm at 15 µg/ml of 2019 QIV–binding clonally unique mAbs generated from singly sorted PBs from week 1 after vaccination. Positive binding is defined as greater than two times the OD 490 value for antibody binding to BSA. (C and D) Flow cytometry gating of total GC B cells (CD20⁺ CD38^int) and HA-binding GC B cells (CD20⁺ CD38^int HA⁺) in the lymph node (LN) from participant 05 (C) and participant 04 (D). Cells were pre-gated on CD4⁻ CD19⁺ IgD⁻ live singlets. (E) OD at 490 nm of 15 µg/ml of 2019 QIV–binding clonally unique mAbs generated from GC B cells at the indicated time points after vaccination. Positive binding defined as greater than two times the OD 490 value for antibody binding to BSA. (F) Unsupervised clustering visualized via UMAP of B cells from blood, LN, and bone marrow (BM) scRNA-seq samples in participant 04. Each dot represents a cell, colored by phenotype as defined by transcriptomic profiles. Naïve B cells (gold), PBs (red), ABCs (green), GC B cells (blue), LNPC (red), RMBs (lavender), and plasma cells (PC, red) populations are pooled from all time points (first panel). QIV-specific cells at each week after vaccination are colored as described. N.D., no data.